{"slug": "gpt-astra", "name": "GPT Astra", "category": "ai", "category_name": "AI models and assistants", "description": "OpenAI's top-tier GPT-6 model, used through ChatGPT/Codex subscription plans and the API.", "recommenders": 47, "commenters": 110, "warners": 20, "score": 37.508, "rising": 45.67, "first": "2026-03-06", "latest": "2026-10-29", "why": "Liked for coding, planning and design review, and for math proofs and CAD or 3D modeling work.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 46, "neutral": 39, "compare": 8, "warn": 18}, "spark": [0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 33, 12], "listed": true, "loved": ["Coding quality", "Planning and review"], "caveat": "Usage limits", "short_quote": "Astra on Extra High. Knowledge work, the right answer is much more important than a fast or cheap answer.", "icon": true, "domain": "gertlabs.com", "github_repo": null, "official_url": "https://gertlabs.com", "pricing": "paid", "aliases": ["Astra", "GPT-6 Astra", "GPT Astra", "Astra 6", "GPT 6 Astra", "GPT-Astra"], "verdict": [{"text": "Liked for coding, planning and design review,", "cites": [{"id": 798371552, "item_id": 49898222, "url": "https://news.ycombinator.com/item?id=49898222", "by": "trentnix", "date": "2026-09-29", "stance": "recommend", "quote": "My experience with Astra (I use it at home writing Go and C) for coding has been fantastic.", "excerpt": "That's not been my experience. My experience with Astra (I use it at home writing Go and C) for coding has been fantastic. Opus 5.5 (I use it for work writing C#) seems faster than Opus 5, but it doesn't seem demonstrably better to my eyes and is still prone to word vomit.", "context": "", "use_case": "coding in Go and C", "switched_from": "", "story": "GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price"}, {"id": 800403121, "item_id": 50025195, "url": "https://news.ycombinator.com/item?id=50025195", "by": "aftbit", "date": "2026-10-09", "stance": "recommend", "quote": "Very tricky stuff, like planning a new feature, design review, or code review of a complex change across multiple repositories is where I leverage Astra.", "excerpt": "I have had good luck rotating between the three tiers of GPT 5.6 with occasional jumps up to Astra or Fable. Most of my work is with GPT-5.6-Sol. Simple tasks like data extraction or trivial refactors (rename this variable etc) I often push down to Terra or Luna, or even to self-hosted Gemma4:31B. Very tricky stuff, like planning a new feature, design review, or code review of a complex change across multiple reposit", "context": "", "use_case": "planning and complex code review", "switched_from": "", "story": "Why isn't the industry freaking out about DeepSeek 4.1 Flash?"}, {"id": 796907472, "item_id": 49806717, "url": "https://news.ycombinator.com/item?id=49806717", "by": "Rapzid", "date": "2026-09-22", "stance": "recommend", "quote": "I use Astra for destroying vaguely scoped asks and tasks, and then for high-level design and plan generations", "excerpt": "Yeah, I use Astra for destroying vaguely scoped asks and tasks, and then for high-level design and plan generations.. Otherwise I'm using 5.6 Sol for actual plan execution and review..", "context": "", "use_case": "planning and design", "switched_from": "", "story": "GPT-6 Sol and Luna"}, {"id": 795549792, "item_id": 49721862, "url": "https://news.ycombinator.com/item?id=49721862", "by": "pod_krad", "date": "2026-09-16", "stance": "recommend", "quote": "Astra xhigh - for specs, analysis.", "excerpt": "Astra xhigh - for specs, analysis. Sol xhigh - for orchestration and implementation in complicated cases. Terra, Luna - for implementation in simple routine cases.", "context": "", "use_case": "specs and analysis", "switched_from": "", "story": "Ask HN: What default model do you use and why?"}]}, {"text": " and for math proofs and CAD or 3D modeling work.", "cites": [{"id": 799543121, "item_id": 49971445, "url": "https://news.ycombinator.com/item?id=49971445", "by": "jychang", "date": "2026-10-05", "stance": "recommend", "quote": "I used OpenAI Dots running GPT-6 Astra on the Ultra reasoning mode", "excerpt": "I used OpenAI Dots running GPT-6 Astra on the Ultra reasoning mode, as an agent swarm with 6 subagents, to create a new mathematical proof. The proof comes with Lean verification. Caveat: this is not peer reviewed yet, so consider it speculative. Still, not bad for a weekend of playing with agent orchestration. I am in touch with a few mathematicians from the Covering Repository, who have confirmed that the result is", "context": "", "use_case": "math proofs", "switched_from": "", "story": "Using OpenAI Dots as an agent swarm to prove new math, for free (Lean verified)"}, {"id": 797576497, "item_id": 49848531, "url": "https://news.ycombinator.com/item?id=49848531", "by": "atemerev", "date": "2026-09-25", "stance": "recommend", "quote": "I used Fable 5.1 and Astra to solve and prove a long-standing mathematical problem", "excerpt": "Just recently I used Fable 5.1 and Astra to solve and prove a long-standing mathematical problem I was always interested in: the shoreline search problem (a ship in total fog is at unknown distance from the shore [infinite line], what is the optimal trajectory?) A particular kind of logarithmic spiral was conjectured 33 years ago; a few days ago, I have obtained the proof. How routine it has become. https://arxiv.org", "context": "mathematician", "use_case": "mathematical proofs", "switched_from": "", "story": "Yes, Claude can do nine loops"}, {"id": 801621840, "item_id": 50101365, "url": "https://news.ycombinator.com/item?id=50101365", "by": "edg5000", "date": "2026-10-15", "stance": "recommend", "quote": "Astra in Codex CLI is good with FreeCAD if you let it use the FreeCAD Python runtime in a headless session.", "excerpt": "Astra in Codex CLI is good with FreeCAD if you let it use the FreeCAD Python runtime in a headless session. It is good at relying purely on natural language to design and revise. To follow specific CAD techniques you need additional prompting. Highly recommend. Uses quite a lot of tokens.", "context": "", "use_case": "designing CAD models via natural language", "switched_from": "", "story": "Ask HN: CAD for wood working with LLM support?"}, {"id": 800024690, "item_id": 50001543, "url": "https://news.ycombinator.com/item?id=50001543", "by": "bitexploder", "date": "2026-10-08", "stance": "recommend", "quote": "Astra is current a bit better than Opus 5.5.", "excerpt": "I have been using these models to make 3d models with build123d \u2014 they are great at it. I am never touching Fusion again. Astra is current a bit better than Opus 5.5.", "context": "", "use_case": "generating 3D models", "switched_from": "", "story": "Claude Haiku 5.5"}]}, {"text": " The most common complaint is that it burns through usage limits fast,", "cites": [{"id": 798274051, "item_id": 49892128, "url": "https://news.ycombinator.com/item?id=49892128", "by": "joshstrange", "date": "2026-09-29", "stance": "warn", "quote": "tried Astra and it also burned though my weekly usage in a day, with a _single_ agent.", "excerpt": "I can believe that (20x is not 4x of the 5x plan) but I\u2019ve been on the $100/mo Codex and the $200/mo Claude plans and I can run the whole week with multiple agents on Claude, mean while I burned through my whole week in 1 day on Sol. Yes, it\u2019s half the price of my Claude plan but even if OpenAi actually gave 4x usage for the next plan up (which they just halved), it would still be less than the usage I get from Claud", "context": "", "use_case": "coding", "switched_from": "", "story": "OpenAI: Tomorrow we are re-opening the Pro $200 subscription"}, {"id": 797999233, "item_id": 49874952, "url": "https://news.ycombinator.com/item?id=49874952", "by": "sitsitstay", "date": "2026-09-28", "stance": "warn", "quote": "I used Astra, and it lasted less than hour before it was dead for a month", "excerpt": "I've been tokenmaxxing across Claude (multiple subs, but I've now burned all of them - and also exhausted, for all of them, the discounted \"Buy Credits\" from 40% to 25% to 20%....). I used Astra, and it lasted less than hour before it was dead for a month - and Grok whilst did solid work, struggled with cohesive architecture. DeepSeek, well, I've got 400+ dollars left in credit I will donate to my enemies, to slow th", "context": "", "use_case": "coding", "switched_from": "", "story": "Open Annihilation \u2013 Spent the GDP of a Small Nation Remaking a Classic 90s Game"}, {"id": 796896819, "item_id": 49806051, "url": "https://news.ycombinator.com/item?id=49806051", "by": "hombre_fatal", "date": "2026-09-22", "stance": "warn", "quote": "Astra blows through usage limits too fast to be a daily driver and writes weird code", "excerpt": "I mainly use Codex/Sol to review my plans drafted by Fable. But beyond that, Astra blows through usage limits too fast to be a daily driver and writes weird code despite what my \"house style\" is, and Codex is behind Claude Code in terms of critical features like seeing what's going on in subagents. The parent + subagent workflow has become critical for keeping the reasoning agent (parent) context-lean while also lett", "context": "", "use_case": "", "switched_from": "", "story": "GPT-6 Sol and Luna"}, {"id": 796647733, "item_id": 49790483, "url": "https://news.ycombinator.com/item?id=49790483", "by": "swalsh", "date": "2026-09-21", "stance": "warn", "quote": "astra does tend to chew up usage", "excerpt": "Codex has become my goto tooling. I used to be a Claude Max subscriber, but I was becoming disappointed with the quality of the output from Opus 5. Fable chewed through my usage too quickly to be practical. Moving to a Pro account w/ Codex was a big improvement. Sol had great output, and the usage was more than sufficient for most of my needs. However astra does tend to chew up usage, so when i've done to much of tha", "context": "", "use_case": "", "switched_from": "", "story": "Grok 4.7"}, {"id": 796950400, "item_id": 49809400, "url": "https://news.ycombinator.com/item?id=49809400", "by": "amluto", "date": "2026-09-22", "stance": "warn", "quote": "I use Astra for rapidly consuming my token limit on a task that would not consume it on 5.6 Sol.", "excerpt": "I use Astra for rapidly consuming my token limit on a task that would not consume it on 5.6 Sol. (I have not done anything quantitative here. For one thing, OpenAI\u2019s billing pages and the codex-rs frontend make it pathetically difficult to get any real data. Some day I should wire up a proxy to extract actual stats.)", "context": "", "use_case": "", "switched_from": "", "story": "GPT-6 Sol and Luna"}]}, {"text": " followed by odd code style and reports of quality degrading.", "cites": [{"id": 802775105, "item_id": 50173444, "url": "https://news.ycombinator.com/item?id=50173444", "by": "LukaD", "date": "2026-10-20", "stance": "warn", "quote": "The last time I used astra it wrote a lot of rust code where it would insert almost no empty lines at all.", "excerpt": "I haven't used prettier much but this is what the prettier docs say about empty lines. > It turns out that empty lines are very hard to automatically generate. The approach that Prettier takes is to preserve empty lines the way they were in the original source code. There are two additional rules: > > - Prettier collapses multiple blank lines into a single blank line. > - Empty lines at the start and end of blocks (a", "context": "", "use_case": "writing Rust code", "switched_from": "", "story": "SlopCodeBench 10/2026 \u2013 Opus 5.5, Astra wins overall, xhigh considered harmful"}, {"id": 796660112, "item_id": 49791257, "url": "https://news.ycombinator.com/item?id=49791257", "by": "brink", "date": "2026-09-21", "stance": "warn", "quote": "Astra is extremely RL fried, and nobody is talking about it.", "excerpt": "Same, Astra is extremely RL fried, and nobody is talking about it. I used Astra for a few days on my personal project, and load times went from less than 3 seconds to almost 30 seconds because it kept using the wrong sync primitives and bad architecture overall.", "context": "", "use_case": "personal project coding", "switched_from": "", "story": "Grok 4.7"}, {"id": 797994051, "item_id": 49874628, "url": "https://news.ycombinator.com/item?id=49874628", "by": "Gareth321", "date": "2026-09-28", "stance": "warn", "quote": "Even Astra has been noticeably degraded for me and everyone else I have asked.", "excerpt": "They're impossible to objectively track by design and intent. Intelligence is such a nebulous target that one can find any number of metrics to support any premise: that the model is smarter, and that the model is dumber. We use benchmarks to attempt to standardise comparisons but these are quickly ingested into the training data and then become effectively useless. You might have heard the term \"bench-maxed.\" Meanin", "context": "", "use_case": "", "switched_from": "", "story": "Ember-1"}, {"id": 795196928, "item_id": 49699808, "url": "https://news.ycombinator.com/item?id=49699808", "by": "varispeed", "date": "2026-09-14", "stance": "warn", "quote": "With Astra and Fable I am no longer able to finish tasks I used to be able to do and I revert to older models.", "excerpt": "The AI has gotten substantially worse. With Astra and Fable I am no longer able to finish tasks I used to be able to do and I revert to older models. Latest models go into reasoning loops, commit logic errors routinely and cannot even spot them by themselves and many more. For instance I have been trying to come up with generalised model of certain non-linear behaviour and had Astra working on it for 4 days. It produ", "context": "", "use_case": "modeling non-linear behaviour", "switched_from": "", "story": "Fable 5.1 Solves the Cyphral Distich, a 370-year-old cipher"}]}], "praise": [{"label": "Coding quality", "summary": "Several commenters rank it near the top for coding, and some found it better than rival models.", "quotes": [{"id": 798371552, "item_id": 49898222, "url": "https://news.ycombinator.com/item?id=49898222", "by": "trentnix", "date": "2026-09-29", "stance": "recommend", "quote": "My experience with Astra (I use it at home writing Go and C) for coding has been fantastic.", "excerpt": "That's not been my experience. My experience with Astra (I use it at home writing Go and C) for coding has been fantastic. Opus 5.5 (I use it for work writing C#) seems faster than Opus 5, but it doesn't seem demonstrably better to my eyes and is still prone to word vomit.", "context": "", "use_case": "coding in Go and C", "switched_from": "", "story": "GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price"}, {"id": 798569553, "item_id": 49910597, "url": "https://news.ycombinator.com/item?id=49910597", "by": "gertlabs", "date": "2026-09-30", "stance": "recommend", "quote": "Both Astra and 6.1-Sol are pretty comfortably ahead of Opus 5.5", "excerpt": "We tested it across 100 unsaturated coding and engineering environments. Both Astra and 6.1-Sol are pretty comfortably ahead of Opus 5.5 in these types of evaluations, and both end up being cheaper than Opus via API usage. 6.1-Sol is also cheaper than Sonnet 5.5 and much smarter. The only verifiable domain Anthropic seems to be clearly ahead is chemistry (and perhaps also some unverifiable domains like being pleasant", "context": "", "use_case": "coding and engineering tasks", "switched_from": "", "story": "GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price"}, {"id": 803494434, "item_id": 50218402, "url": "https://news.ycombinator.com/item?id=50218402", "by": "sail0rm00n", "date": "2026-10-24", "stance": "recommend", "quote": "2. Astra", "excerpt": "As tribute, I bit the bullet on the $40 plan to test out K3.1 Pro. Still burning through the 5-hour rate limit way too quickly, though. Output quality seems solid enough, but the value will probably make a lot more sense via API pricing for smaller projects. My current tier list based on what I\u2019ve been testing: 1. Opus 5.5 2. Astra 3. Argon 4. Kimi K3.1 Pro (For lightweight/daily driver tasks, Gemini Flash 3.8, DeepS", "context": "", "use_case": "coding", "switched_from": "", "story": "Kimi K3.1 Pro and K3.1 Pro Highspeed"}]}, {"label": "Planning and review", "summary": "Often reserved for specs, high-level design, planning and code review in multi-model setups.", "quotes": [{"id": 800403121, "item_id": 50025195, "url": "https://news.ycombinator.com/item?id=50025195", "by": "aftbit", "date": "2026-10-09", "stance": "recommend", "quote": "Very tricky stuff, like planning a new feature, design review, or code review of a complex change across multiple repositories is where I leverage Astra.", "excerpt": "I have had good luck rotating between the three tiers of GPT 5.6 with occasional jumps up to Astra or Fable. Most of my work is with GPT-5.6-Sol. Simple tasks like data extraction or trivial refactors (rename this variable etc) I often push down to Terra or Luna, or even to self-hosted Gemma4:31B. Very tricky stuff, like planning a new feature, design review, or code review of a complex change across multiple reposit", "context": "", "use_case": "planning and complex code review", "switched_from": "", "story": "Why isn't the industry freaking out about DeepSeek 4.1 Flash?"}, {"id": 796907472, "item_id": 49806717, "url": "https://news.ycombinator.com/item?id=49806717", "by": "Rapzid", "date": "2026-09-22", "stance": "recommend", "quote": "I use Astra for destroying vaguely scoped asks and tasks, and then for high-level design and plan generations", "excerpt": "Yeah, I use Astra for destroying vaguely scoped asks and tasks, and then for high-level design and plan generations.. Otherwise I'm using 5.6 Sol for actual plan execution and review..", "context": "", "use_case": "planning and design", "switched_from": "", "story": "GPT-6 Sol and Luna"}, {"id": 795263488, "item_id": 49703968, "url": "https://news.ycombinator.com/item?id=49703968", "by": "dlahoda", "date": "2026-09-14", "stance": "recommend", "quote": "burns half of day sub of astra for 200 USD. runs 1 hour on our repo.", "excerpt": "I would and do. works perfect https://github.com/dzmitry-lahoda/dz/tree/main/agents/skills/code-review burns half of day sub of astra for 200 USD. runs 1 hour on our repo. finds bugs missed by coderabitai, devin-integration-bot, codex and copilot (I ask them first until nothing found, yet my orchestration finds more). I do not hardcode our application in orchestration, but run subagent for applicationdomainproduct de", "context": "", "use_case": "code review orchestration", "switched_from": "", "story": "GPT-5.6 Luna vs. GPT-6 Astra: Is a $1.20 Model Good Enough for Code Review?"}]}, {"label": "CAD and 3D", "summary": "Reported to work well for CAD, 3D modeling and design plans through tools like FreeCAD, build123d and Blender MCP.", "quotes": [{"id": 801621840, "item_id": 50101365, "url": "https://news.ycombinator.com/item?id=50101365", "by": "edg5000", "date": "2026-10-15", "stance": "recommend", "quote": "Astra in Codex CLI is good with FreeCAD if you let it use the FreeCAD Python runtime in a headless session.", "excerpt": "Astra in Codex CLI is good with FreeCAD if you let it use the FreeCAD Python runtime in a headless session. It is good at relying purely on natural language to design and revise. To follow specific CAD techniques you need additional prompting. Highly recommend. Uses quite a lot of tokens.", "context": "", "use_case": "designing CAD models via natural language", "switched_from": "", "story": "Ask HN: CAD for wood working with LLM support?"}, {"id": 800024690, "item_id": 50001543, "url": "https://news.ycombinator.com/item?id=50001543", "by": "bitexploder", "date": "2026-10-08", "stance": "recommend", "quote": "Astra is current a bit better than Opus 5.5.", "excerpt": "I have been using these models to make 3d models with build123d \u2014 they are great at it. I am never touching Fusion again. Astra is current a bit better than Opus 5.5.", "context": "", "use_case": "generating 3D models", "switched_from": "", "story": "Claude Haiku 5.5"}, {"id": 801577361, "item_id": 50098585, "url": "https://news.ycombinator.com/item?id=50098585", "by": "greenpizza13", "date": "2026-10-15", "stance": "recommend", "quote": "I used blender mcp and Astra to do a woodworking design a few days ago, plans came out great.", "excerpt": "I used blender mcp and Astra to do a woodworking design a few days ago, plans came out great.", "context": "", "use_case": "woodworking design", "switched_from": "", "story": "Ask HN: CAD for wood working with LLM support?"}]}, {"label": "Math and research", "summary": "Commenters report using it to produce new mathematical proofs and for knowledge work where accuracy matters.", "quotes": [{"id": 799543121, "item_id": 49971445, "url": "https://news.ycombinator.com/item?id=49971445", "by": "jychang", "date": "2026-10-05", "stance": "recommend", "quote": "I used OpenAI Dots running GPT-6 Astra on the Ultra reasoning mode", "excerpt": "I used OpenAI Dots running GPT-6 Astra on the Ultra reasoning mode, as an agent swarm with 6 subagents, to create a new mathematical proof. The proof comes with Lean verification. Caveat: this is not peer reviewed yet, so consider it speculative. Still, not bad for a weekend of playing with agent orchestration. I am in touch with a few mathematicians from the Covering Repository, who have confirmed that the result is", "context": "", "use_case": "math proofs", "switched_from": "", "story": "Using OpenAI Dots as an agent swarm to prove new math, for free (Lean verified)"}, {"id": 797576497, "item_id": 49848531, "url": "https://news.ycombinator.com/item?id=49848531", "by": "atemerev", "date": "2026-09-25", "stance": "recommend", "quote": "I used Fable 5.1 and Astra to solve and prove a long-standing mathematical problem", "excerpt": "Just recently I used Fable 5.1 and Astra to solve and prove a long-standing mathematical problem I was always interested in: the shoreline search problem (a ship in total fog is at unknown distance from the shore [infinite line], what is the optimal trajectory?) A particular kind of logarithmic spiral was conjectured 33 years ago; a few days ago, I have obtained the proof. How routine it has become. https://arxiv.org", "context": "mathematician", "use_case": "mathematical proofs", "switched_from": "", "story": "Yes, Claude can do nine loops"}, {"id": 797686672, "item_id": 49855417, "url": "https://news.ycombinator.com/item?id=49855417", "by": "LUmBULtERA", "date": "2026-09-26", "stance": "recommend", "quote": "Astra on Extra High. Knowledge work, the right answer is much more important than a fast or cheap answer.", "excerpt": "Astra on Extra High. Knowledge work, the right answer is much more important than a fast or cheap answer. If I were to switch to Anthropic, Opus 5.5 or Fable 5.1 on high or better.", "context": "", "use_case": "knowledge work", "switched_from": "", "story": "Ask HN: Which model do you use for work?"}]}, {"label": "Instruction following", "summary": "Commenters say it follows AGENTS.md instructions closely and does not cheat, for example by deleting failing tests.", "quotes": [{"id": 803487105, "item_id": 50217944, "url": "https://news.ycombinator.com/item?id=50217944", "by": "RussianCow", "date": "2026-10-24", "stance": "recommend", "quote": "Sol and Astra are by far the best of the models I use at following the instructions in AGENTS.md", "excerpt": "This is my experience as well. Sol and Astra are by far the best of the models I use at following the instructions in AGENTS.md, almost to an annoying point. (But I'd rather have that problem than the inverse.)", "context": "", "use_case": "following agent instructions", "switched_from": "", "story": "Kimi K3.1 Pro and K3.1 Pro Highspeed"}, {"id": 795029936, "item_id": 49689371, "url": "https://news.ycombinator.com/item?id=49689371", "by": "ainch", "date": "2026-09-13", "stance": "recommend", "quote": "I use Astra/Sol for coding, and I haven't run into any issues with them trying to hack someone or break the law", "excerpt": "This is true in principle. But at the same time, I use Astra/Sol for coding, and I haven't run into any issues with them trying to hack someone or break the law when they reach an impasse. Not even more trivial things like deleting non-passing tests. I find it hard to reconcile. I wonder to what degree this is because models can identify that they are in graded/eval environments, and therefore conclude that there are", "context": "", "use_case": "coding", "switched_from": "", "story": "Astra and Fable still hack on simple variants of alignment evals from 2025"}]}, {"label": "Concise output", "summary": "One former Claude user found its answers more concise and information-dense while being similarly accurate.", "quotes": [{"id": 794809584, "item_id": 49675599, "url": "https://news.ycombinator.com/item?id=49675599", "by": "asd88", "date": "2026-09-12", "stance": "recommend", "quote": "Astra\u2019s outputs are much more concise while being similarly accurate, they seem more information dense.", "excerpt": "I have \u201cunlimited\u201d (but metered) tokens at work. I was primarily a Claude (Opus and Fable) user until Astra came out. Astra\u2019s outputs are much more concise while being similarly accurate, they seem more information dense. It\u2019s also extremely fast and doesn\u2019t have to \u201cgo check instead of answering from memory\u201d if the answer is already in context, or dig super deep into a project before answering. It\u2019s honestly refresh", "context": "has unlimited metered tokens at work", "use_case": "professional software engineering", "switched_from": "Claude", "story": "Ask HN: What default model do you use and why?"}]}], "caveats": [{"label": "Usage limits", "summary": "Many say it exhausts weekly or subscription usage very quickly, and some call it too expensive.", "quotes": [{"id": 798274051, "item_id": 49892128, "url": "https://news.ycombinator.com/item?id=49892128", "by": "joshstrange", "date": "2026-09-29", "stance": "warn", "quote": "tried Astra and it also burned though my weekly usage in a day, with a _single_ agent.", "excerpt": "I can believe that (20x is not 4x of the 5x plan) but I\u2019ve been on the $100/mo Codex and the $200/mo Claude plans and I can run the whole week with multiple agents on Claude, mean while I burned through my whole week in 1 day on Sol. Yes, it\u2019s half the price of my Claude plan but even if OpenAi actually gave 4x usage for the next plan up (which they just halved), it would still be less than the usage I get from Claud", "context": "", "use_case": "coding", "switched_from": "", "story": "OpenAI: Tomorrow we are re-opening the Pro $200 subscription"}, {"id": 797999233, "item_id": 49874952, "url": "https://news.ycombinator.com/item?id=49874952", "by": "sitsitstay", "date": "2026-09-28", "stance": "warn", "quote": "I used Astra, and it lasted less than hour before it was dead for a month", "excerpt": "I've been tokenmaxxing across Claude (multiple subs, but I've now burned all of them - and also exhausted, for all of them, the discounted \"Buy Credits\" from 40% to 25% to 20%....). I used Astra, and it lasted less than hour before it was dead for a month - and Grok whilst did solid work, struggled with cohesive architecture. DeepSeek, well, I've got 400+ dollars left in credit I will donate to my enemies, to slow th", "context": "", "use_case": "coding", "switched_from": "", "story": "Open Annihilation \u2013 Spent the GDP of a Small Nation Remaking a Classic 90s Game"}, {"id": 796896819, "item_id": 49806051, "url": "https://news.ycombinator.com/item?id=49806051", "by": "hombre_fatal", "date": "2026-09-22", "stance": "warn", "quote": "Astra blows through usage limits too fast to be a daily driver and writes weird code", "excerpt": "I mainly use Codex/Sol to review my plans drafted by Fable. But beyond that, Astra blows through usage limits too fast to be a daily driver and writes weird code despite what my \"house style\" is, and Codex is behind Claude Code in terms of critical features like seeing what's going on in subagents. The parent + subagent workflow has become critical for keeping the reasoning agent (parent) context-lean while also lett", "context": "", "use_case": "", "switched_from": "", "story": "GPT-6 Sol and Luna"}]}, {"label": "Code style", "summary": "Commenters report odd code such as missing blank lines, overly complex functions, and poor architectural choices.", "quotes": [{"id": 802775105, "item_id": 50173444, "url": "https://news.ycombinator.com/item?id=50173444", "by": "LukaD", "date": "2026-10-20", "stance": "warn", "quote": "The last time I used astra it wrote a lot of rust code where it would insert almost no empty lines at all.", "excerpt": "I haven't used prettier much but this is what the prettier docs say about empty lines. > It turns out that empty lines are very hard to automatically generate. The approach that Prettier takes is to preserve empty lines the way they were in the original source code. There are two additional rules: > > - Prettier collapses multiple blank lines into a single blank line. > - Empty lines at the start and end of blocks (a", "context": "", "use_case": "writing Rust code", "switched_from": "", "story": "SlopCodeBench 10/2026 \u2013 Opus 5.5, Astra wins overall, xhigh considered harmful"}, {"id": 802789092, "item_id": 50174318, "url": "https://news.ycombinator.com/item?id=50174318", "by": "throwaway219450", "date": "2026-10-20", "stance": "warn", "quote": "Astra also creates things like \u201creport\u201d functions with 10 arguments", "excerpt": "I was talking about a function with 100 lines and no line breaks. Very difficult to read. I use black/ruff on a lot of projects and it will only touch standard-y things like docstring spacing, or spacing between methods and classes, import sorting etc. It doesn\u2019t touch intentional/aesthetic user spacing. With agents it is helpful to make them run ruff and mypy to keep them honest, but as long as the code adheres to t", "context": "", "use_case": "AI coding", "switched_from": "", "story": "SlopCodeBench 10/2026 \u2013 Opus 5.5, Astra wins overall, xhigh considered harmful"}, {"id": 796660112, "item_id": 49791257, "url": "https://news.ycombinator.com/item?id=49791257", "by": "brink", "date": "2026-09-21", "stance": "warn", "quote": "Astra is extremely RL fried, and nobody is talking about it.", "excerpt": "Same, Astra is extremely RL fried, and nobody is talking about it. I used Astra for a few days on my personal project, and load times went from less than 3 seconds to almost 30 seconds because it kept using the wrong sync primitives and bad architecture overall.", "context": "", "use_case": "personal project coding", "switched_from": "", "story": "Grok 4.7"}]}, {"label": "Perceived degradation", "summary": "Some users believe it was silently downgraded or became worse after launch.", "quotes": [{"id": 797994051, "item_id": 49874628, "url": "https://news.ycombinator.com/item?id=49874628", "by": "Gareth321", "date": "2026-09-28", "stance": "warn", "quote": "Even Astra has been noticeably degraded for me and everyone else I have asked.", "excerpt": "They're impossible to objectively track by design and intent. Intelligence is such a nebulous target that one can find any number of metrics to support any premise: that the model is smarter, and that the model is dumber. We use benchmarks to attempt to standardise comparisons but these are quickly ingested into the training data and then become effectively useless. You might have heard the term \"bench-maxed.\" Meanin", "context": "", "use_case": "", "switched_from": "", "story": "Ember-1"}, {"id": 796774720, "item_id": 49798420, "url": "https://news.ycombinator.com/item?id=49798420", "by": "Gareth321", "date": "2026-09-22", "stance": "warn", "quote": "Many empirical examples of Astra silently downgrading the models.", "excerpt": "I strongly agree. Check out the Codex subreddit. Many empirical examples of Astra silently downgrading the models. One found Astra was silently using Luna Max (but still billing for Astra). Even when I try to stick with Sol X/High, my limits are at best half of what they were before Astra launched, and the intelligence has declined markedly. I cancelled my $100 plan. This is absolutely absurd and frankly unusable now", "context": "paying Codex subscriber", "use_case": "coding", "switched_from": "", "story": "MiMo-v2.6-Pro: Intelligence, Performance and Price Analysis"}, {"id": 795196928, "item_id": 49699808, "url": "https://news.ycombinator.com/item?id=49699808", "by": "varispeed", "date": "2026-09-14", "stance": "warn", "quote": "With Astra and Fable I am no longer able to finish tasks I used to be able to do and I revert to older models.", "excerpt": "The AI has gotten substantially worse. With Astra and Fable I am no longer able to finish tasks I used to be able to do and I revert to older models. Latest models go into reasoning loops, commit logic errors routinely and cannot even spot them by themselves and many more. For instance I have been trying to come up with generalised model of certain non-linear behaviour and had Astra working on it for 4 days. It produ", "context": "", "use_case": "modeling non-linear behaviour", "switched_from": "", "story": "Fable 5.1 Solves the Cyphral Distich, a 370-year-old cipher"}]}, {"label": "Fails hard tasks", "summary": "Commenters report failures on a large port, PCB schematics, and test suites that missed bugs.", "quotes": [{"id": 800055937, "item_id": 50003496, "url": "https://news.ycombinator.com/item?id=50003496", "by": "flossly", "date": "2026-10-08", "stance": "warn", "quote": "never got past like 84% compat.", "excerpt": "I found the whole section super interesting... I'll copy it here: I used a lot of OpenAI models to try and complete this port. In total I did over $400,000 in API priced tokens with GPT-5.6 Sol and GPT 6 Astra. They wrote over 1.3m lines of Rust over multiple months of /goal loops and never got past like 84% compat. When I saw how little my Claude Code limits were burning, I figured it'd be fun to throw Opus 5.5 at t", "context": "", "use_case": "porting TypeScript compiler to Rust", "switched_from": "", "story": "Port of the TypeScript compiler, checker and lsp to Rust, by LLM"}, {"id": 795840129, "item_id": 49740008, "url": "https://news.ycombinator.com/item?id=49740008", "by": "the__alchemist", "date": "2026-09-17", "stance": "warn", "quote": "Those models were not capable of even schematic editing or building.", "excerpt": "Hey - I ordered a board yesterday. I used Opus 5, Astra, and Sol at points to QC the board, and offer power routing advice. (The MCU is a brand new one, and poorly documented unless they send you an NDA and the full docs). Those models were not capable of even schematic editing or building. (I didn't bother with routing or placement). They made gross errors all over the schematic; changed the known-working parts and ", "context": "", "use_case": "PCB schematic QC", "switched_from": "", "story": "This PCB is brought to you by Fable 5"}, {"id": 795458337, "item_id": 49716146, "url": "https://news.ycombinator.com/item?id=49716146", "by": "unicornvom12", "date": "2026-09-15", "stance": "warn", "quote": "64 of the bugs that abloh planted had passed Astra's test suite", "excerpt": "This morning we used 'AGI' to build 3 features for our CRM. We then ran abloh.dev on the PR and found that 64 of the bugs that abloh planted had passed Astra's test suite. This is what slipped past Astra and the (avoided) consequences: 1. A blind spot only abloh's AI found The test data Astra used gives every fake company a name identical to its web domain. When abloh's AI swapped name and domain in Astra's query, al", "context": "", "use_case": "building CRM features", "switched_from": "", "story": "Verification Layer for Any Model"}]}, {"label": "Slow, rabbit-holes", "summary": "Some found it very slow on high reasoning settings and prone to chasing minor implementation details.", "quotes": [{"id": 798388800, "item_id": 49899300, "url": "https://news.ycombinator.com/item?id=49899300", "by": "moshegramovsky", "date": "2026-09-29", "stance": "warn", "quote": "I'm not going to keep using Astra unless I can get some benefit from it", "excerpt": "100% hard agree. I used about 10 hours of Astra high-thinking compute time and it was a bad experience. Incredibly slow (prompts running for 30/40 minutes) to do simple things. As a result, Astra didn't get much done. It needs the same small implementation slices as GPT 5.5/others, but was much slower and didn't generate better results. (On a complex infra project/across a large codebase.) It was absolutely terrible ", "context": "pays $200/month for OpenAI", "use_case": "large codebase coding", "switched_from": "", "story": "GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price"}, {"id": 796933586, "item_id": 49808349, "url": "https://news.ycombinator.com/item?id=49808349", "by": "spockz", "date": "2026-09-22", "stance": "warn", "quote": "Whereas Sol and Astra kept rabbit holing and looking for tiny implementation errors.", "excerpt": "I use it for golang, and it is fantastic. Incredibly fast. It seems the llm and I \u201cunderstand\u201d each other. I have to be less careful in my exact phrasing. It kind of just does what I want and expect. When I ask for an explanation it adds the right amount of detail. Of course, some of the material is new to me so subtle errors are hard to spot. But at least I\u2019ve caught Terra and Sol on inconsistent messaging. Also I\u2019v", "context": "", "use_case": "", "switched_from": "", "story": "GPT-6 Sol and Luna"}]}], "who": [{"by": "boogiefromzk", "context": "built an agentic delegation scheme", "url": "https://news.ycombinator.com/item?id=50251451", "date": "2026-10-26"}, {"by": "potsandpans", "context": "decompiling a PlayStation game", "url": "https://news.ycombinator.com/item?id=50222286", "date": "2026-10-24"}, {"by": "creilly11235", "context": "author solving unsolved cryptograms", "url": "https://news.ycombinator.com/item?id=50167285", "date": "2026-10-20"}, {"by": "Livia_Zaharia", "context": "architect, jewelry designer, startup founder", "url": "https://news.ycombinator.com/item?id=49964234", "date": "2026-10-05"}, {"by": "johnsmith1840", "context": "has around $500 a month AI budget", "url": "https://news.ycombinator.com/item?id=49898755", "date": "2026-09-29"}, {"by": "atemerev", "context": "mathematician", "url": "https://news.ycombinator.com/item?id=49848531", "date": "2026-09-25"}, {"by": "jimmaswell", "context": "game developer", "url": "https://news.ycombinator.com/item?id=49817549", "date": "2026-09-23"}, {"by": "matheusmoreira", "context": "former Anthropic customer", "url": "https://news.ycombinator.com/item?id=49809843", "date": "2026-09-22"}, {"by": "pkulak", "context": "$20 subscription user", "url": "https://news.ycombinator.com/item?id=49706632", "date": "2026-09-15"}, {"by": "jakevoytko", "context": "former Google Docs team member", "url": "https://news.ycombinator.com/item?id=49697215", "date": "2026-09-14"}, {"by": "asd88", "context": "has unlimited metered tokens at work", "url": "https://news.ycombinator.com/item?id=49675599", "date": "2026-09-12"}, {"by": "fibonacci112358", "context": "on a 20x plan", "url": "https://news.ycombinator.com/item?id=49673841", "date": "2026-09-12"}], "tags": ["for planning and code review", "for CAD and 3D modeling", "for math research", "for Codex CLI users", "for multi-model agent workflows"], "use_cases": ["coding", "coding in codex", "code review", "coding agent", "general-purpose ai tasks", "moonshot tasks"], "compared": [{"slug": "gpt-sol", "name": "GPT Sol", "co": 45, "switched_from": 2, "switched_to": 2}, {"slug": "claude-opus", "name": "Claude Opus", "co": 41, "switched_from": 1, "switched_to": 1}, {"slug": "claude-fable", "name": "Claude Fable", "co": 31, "switched_from": 2, "switched_to": 1}, {"slug": "codex", "name": "Codex", "co": 26, "switched_from": 0, "switched_to": 0}, {"slug": "gpt-luna", "name": "GPT Luna", "co": 23, "switched_from": 0, "switched_to": 0}, {"slug": "claude-code", "name": "Claude Code", "co": 9, "switched_from": 1, "switched_to": 0}, {"slug": "claude", "name": "Claude", "co": 7, "switched_from": 2, "switched_to": 0}, {"slug": "gpt-terra", "name": "GPT Terra", "co": 9, "switched_from": 0, "switched_to": 0}], "makers": [{"id": 756495280, "item_id": 47280955, "url": "https://news.ycombinator.com/item?id=47280955", "by": "pppone", "date": "2026-03-06", "stance": "recommend", "quote": "All built in, including autoguiding and autofocusing.", "excerpt": "No MaximDL needed! All built in, including autoguiding and autofocusing. Satellite tracking to be implemented in the future, working on it these months with a student of mine. Object scheduling timing is down to the user - but the wait directives does sound like a useful feature! Custom python scripts are possible, please check out here: https://docs.withastra.io/user_guide/custom_observatories I also recommend check", "context": "", "use_case": "observatory control", "switched_from": "", "story": "Astra: An open-source observatory control software"}], "threads": [{"url": "https://news.ycombinator.com/item?id=49672966", "title": "Ask HN: What default model do you use and why?", "mentions": 11}, {"url": "https://news.ycombinator.com/item?id=49703003", "title": "GPT-5.6 Luna vs. GPT-6 Astra: Is a $1.20 Model Good Enough for Code Review?", "mentions": 8}, {"url": "https://news.ycombinator.com/item?id=49805509", "title": "GPT-6 Sol and Luna", "mentions": 6}, {"url": "https://news.ycombinator.com/item?id=49896586", "title": "GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price", "mentions": 4}, {"url": "https://news.ycombinator.com/item?id=50108717", "title": "SlopCodeBench 10/2026 \u2013 Opus 5.5, Astra wins overall, xhigh considered harmful", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=50000488", "title": "Why isn't the industry freaking out about DeepSeek 4.1 Flash?", "mentions": 3}], "sentiment_by_month": {"2025-11": {}, "2025-12": {}, "2026-01": {}, "2026-02": {}, "2026-03": {"recommend": 1}, "2026-04": {}, "2026-05": {}, "2026-06": {}, "2026-07": {}, "2026-08": {}, "2026-09": {"neutral": 32, "recommend": 33, "compare": 5, "warn": 15}, "2026-10": {"compare": 3, "neutral": 7, "recommend": 12, "warn": 3}}, "quotes": [{"id": 804634403, "item_id": 50289650, "url": "https://news.ycombinator.com/item?id=50289650", "by": "dichiara19", "date": "2026-10-29", "stance": "compare", "quote": "I don\u2019t think it measures up to Fable 5.1 or Astra", "excerpt": "I want to talk about my experience with Argon, noting that I used it via OpenRouter on a Raspberry Pi, spending around $15 in credit. My experience is that Google is aiming to create a generalist model \u2013 but a truly generalist one! In fact, I find it particularly accurate in general terms and even in long conversations, maintaining excellent consistency and a wealth of information in response to complex questions. I\u2019", "context": "", "use_case": "coding agent", "switched_from": "", "story": "Gemini 4 Carbon is the new #1 on the Artificial Analysis Intelligence Index v5"}, {"id": 804047312, "item_id": 50252957, "url": "https://news.ycombinator.com/item?id=50252957", "by": "scarab92", "date": "2026-10-27", "stance": "recommend", "quote": "I use Astra 6 Extra High for everything.", "excerpt": "I use Astra 6 Extra High for everything. It\u2019s overkill most of the time, but frankly all these models are \u201ccheap\u201d, and I\u2019d rather waste compute than brain cycles trying to discern which dials are best for each individual as-hoc query.", "context": "", "use_case": "general-purpose AI tasks", "switched_from": "", "story": "GPT-6.1 Terra"}, {"id": 804023222, "item_id": 50251451, "url": "https://news.ycombinator.com/item?id=50251451", "by": "boogiefromzk", "date": "2026-10-26", "stance": "neutral", "quote": "Also should work in as pair of Astra + Luna in CodeX.", "excerpt": "Last months I was searching for a way to make smart AI model submit and verify the work to cheap model so that it will cost less, and after many try and failures I was able to get stable result where I use Fable and its limits spent slower than overall account limit because most context intensive work is done by Sonnet. At the same time the response quaility is still Fable. I also made sure Opus controls Grok in Curs", "context": "built an agentic delegation scheme", "use_case": "", "switched_from": "", "story": "Cost-effective agentic delegation scheme for Claude Code, Cursor and Codex"}, {"id": 803989666, "item_id": 50249354, "url": "https://news.ycombinator.com/item?id=50249354", "by": "elxr", "date": "2026-10-26", "stance": "neutral", "quote": "Astra is really not relevant for the average dev except to do some moonshot task", "excerpt": "If you daily all these models (and have a modest budget, not some >$400 corporate allowance), then it's pretty obvious that Luna is small and fast, and Sol 6.1 is the big one for jobs that need significantly above average intelligence, comparatively to a human engineer. I use both constantly. Astra is really not relevant for the average dev except to do some moonshot task with a vague verification loop. Sure Sol 6.1 ", "context": "", "use_case": "moonshot tasks", "switched_from": "", "story": "GPT-6.1 Terra"}, {"id": 803899731, "item_id": 50243733, "url": "https://news.ycombinator.com/item?id=50243733", "by": "Sphax", "date": "2026-10-26", "stance": "neutral", "quote": "when they're not I use Astra or Opus 5.5", "excerpt": "I use Opus 5.5 daily at work and a mix of GLM 5.3 (Flash or not), Deepseek v4 (flash or pro) at home and while Opus 5.5 is more capable, in my opinion it matters for less than 5% of the work i do. These chinese models are simply good enough for the vast majority of tasks I run and when they're not I use Astra or Opus 5.5. Yes, leaving Opus 5.5 on all the time is comfortable when you pay the 90\u20ac plan but it's just not", "context": "", "use_case": "fallback for hard tasks", "switched_from": "", "story": "Anthropic S-1"}, {"id": 803556577, "item_id": 50222286, "url": "https://news.ycombinator.com/item?id=50222286", "by": "potsandpans", "date": "2026-10-24", "stance": "recommend", "quote": "Opus and Astra take them handily.", "excerpt": "I recently just crossed the threshold for 100% decompilation of a PlayStation game. I used a lot of models in the process, and I feel like this is a good demonstration the \"sauce\" that opus has. The harness I've built for it is pretty straight forward: split binary into individual functions, give the agent a bunch of small tools to diff output with underlying binary, make them wrote clean c until there's a certified ", "context": "decompiling a PlayStation game", "use_case": "decompiling hard functions", "switched_from": "", "story": "Kimi K3.1 scores 55 on Artificial Analysis, placed on the Pareto line"}, {"id": 803494434, "item_id": 50218402, "url": "https://news.ycombinator.com/item?id=50218402", "by": "sail0rm00n", "date": "2026-10-24", "stance": "recommend", "quote": "2. Astra", "excerpt": "As tribute, I bit the bullet on the $40 plan to test out K3.1 Pro. Still burning through the 5-hour rate limit way too quickly, though. Output quality seems solid enough, but the value will probably make a lot more sense via API pricing for smaller projects. My current tier list based on what I\u2019ve been testing: 1. Opus 5.5 2. Astra 3. Argon 4. Kimi K3.1 Pro (For lightweight/daily driver tasks, Gemini Flash 3.8, DeepS", "context": "", "use_case": "coding", "switched_from": "", "story": "Kimi K3.1 Pro and K3.1 Pro Highspeed"}, {"id": 803487105, "item_id": 50217944, "url": "https://news.ycombinator.com/item?id=50217944", "by": "RussianCow", "date": "2026-10-24", "stance": "recommend", "quote": "Sol and Astra are by far the best of the models I use at following the instructions in AGENTS.md", "excerpt": "This is my experience as well. Sol and Astra are by far the best of the models I use at following the instructions in AGENTS.md, almost to an annoying point. (But I'd rather have that problem than the inverse.)", "context": "", "use_case": "following agent instructions", "switched_from": "", "story": "Kimi K3.1 Pro and K3.1 Pro Highspeed"}, {"id": 802789092, "item_id": 50174318, "url": "https://news.ycombinator.com/item?id=50174318", "by": "throwaway219450", "date": "2026-10-20", "stance": "warn", "quote": "Astra also creates things like \u201creport\u201d functions with 10 arguments", "excerpt": "I was talking about a function with 100 lines and no line breaks. Very difficult to read. I use black/ruff on a lot of projects and it will only touch standard-y things like docstring spacing, or spacing between methods and classes, import sorting etc. It doesn\u2019t touch intentional/aesthetic user spacing. With agents it is helpful to make them run ruff and mypy to keep them honest, but as long as the code adheres to t", "context": "", "use_case": "AI coding", "switched_from": "", "story": "SlopCodeBench 10/2026 \u2013 Opus 5.5, Astra wins overall, xhigh considered harmful"}, {"id": 802775105, "item_id": 50173444, "url": "https://news.ycombinator.com/item?id=50173444", "by": "LukaD", "date": "2026-10-20", "stance": "warn", "quote": "The last time I used astra it wrote a lot of rust code where it would insert almost no empty lines at all.", "excerpt": "I haven't used prettier much but this is what the prettier docs say about empty lines. > It turns out that empty lines are very hard to automatically generate. The approach that Prettier takes is to preserve empty lines the way they were in the original source code. There are two additional rules: > > - Prettier collapses multiple blank lines into a single blank line. > - Empty lines at the start and end of blocks (a", "context": "", "use_case": "writing Rust code", "switched_from": "", "story": "SlopCodeBench 10/2026 \u2013 Opus 5.5, Astra wins overall, xhigh considered harmful"}, {"id": 802763586, "item_id": 50172724, "url": "https://news.ycombinator.com/item?id=50172724", "by": "CharlieDigital", "date": "2026-10-20", "stance": "recommend", "quote": "\"End-User Customer Advocate\" -> Astra, plan mode.", "excerpt": "I use the frontier models only in plan mode; coding is held out to cheaper models. I run Paseo as an orchestrator with agent profiles. \"Distinguished Member of Technical Staff\" -> Fable, plan mode. \"End-User Customer Advocate\" -> Astra, plan mode. A few others. Basically, these agents (with profile prompts) can only be \"consulted\" for input on design elements; it's OK if they produce a bit of verbosity here. This has", "context": "", "use_case": "planning/design", "switched_from": "", "story": "SlopCodeBench 10/2026 \u2013 Opus 5.5, Astra wins overall, xhigh considered harmful"}, {"id": 802676561, "item_id": 50167285, "url": "https://news.ycombinator.com/item?id=50167285", "by": "creilly11235", "date": "2026-10-20", "stance": "neutral", "quote": "Astra verified the results", "excerpt": "Author here, this was the last of four cryptograms from #35 of Klaus Schmeh\u2019s list of 50 unsolved cryptograms. The first 3 were cracked in 1948, 1996 and 2019 with the last one remaining uncracked for 76 years. Wood initially set up the cipher to try to prove he could pass the key through a medium after his death. The key ended up being the Lord's Prayer from a 1912 German Bible. The message reads Hier bin ich. Totsi", "context": "author solving unsolved cryptograms", "use_case": "verifying cryptogram results", "switched_from": "", "story": "Solving the 1950 cipher meant to test life after death"}, {"id": 801621840, "item_id": 50101365, "url": "https://news.ycombinator.com/item?id=50101365", "by": "edg5000", "date": "2026-10-15", "stance": "recommend", "quote": "Astra in Codex CLI is good with FreeCAD if you let it use the FreeCAD Python runtime in a headless session.", "excerpt": "Astra in Codex CLI is good with FreeCAD if you let it use the FreeCAD Python runtime in a headless session. It is good at relying purely on natural language to design and revise. To follow specific CAD techniques you need additional prompting. Highly recommend. Uses quite a lot of tokens.", "context": "", "use_case": "designing CAD models via natural language", "switched_from": "", "story": "Ask HN: CAD for wood working with LLM support?"}, {"id": 801577361, "item_id": 50098585, "url": "https://news.ycombinator.com/item?id=50098585", "by": "greenpizza13", "date": "2026-10-15", "stance": "recommend", "quote": "I used blender mcp and Astra to do a woodworking design a few days ago, plans came out great.", "excerpt": "I used blender mcp and Astra to do a woodworking design a few days ago, plans came out great.", "context": "", "use_case": "woodworking design", "switched_from": "", "story": "Ask HN: CAD for wood working with LLM support?"}, {"id": 800420801, "item_id": 50026300, "url": "https://news.ycombinator.com/item?id=50026300", "by": "mstank", "date": "2026-10-09", "stance": "recommend", "quote": "the latest models -- Opus 5.5, Astra, etc. juggle multiple tasks, delegate exceptionally well and are very good at working independently.", "excerpt": "I used to relate to this article quite a bit. In the last 3-4 months, not so much. I've found that the latest models -- Opus 5.5, Astra, etc. juggle multiple tasks, delegate exceptionally well and are very good at working independently. I still occasionally have issues with open-weight models, but the frontier labs have solved the above for most use cases.", "context": "", "use_case": "agentic multi-task work", "switched_from": "", "story": "Why are coding agents so dumb?"}, {"id": 800403121, "item_id": 50025195, "url": "https://news.ycombinator.com/item?id=50025195", "by": "aftbit", "date": "2026-10-09", "stance": "recommend", "quote": "Very tricky stuff, like planning a new feature, design review, or code review of a complex change across multiple repositories is where I leverage Astra.", "excerpt": "I have had good luck rotating between the three tiers of GPT 5.6 with occasional jumps up to Astra or Fable. Most of my work is with GPT-5.6-Sol. Simple tasks like data extraction or trivial refactors (rename this variable etc) I often push down to Terra or Luna, or even to self-hosted Gemma4:31B. Very tricky stuff, like planning a new feature, design review, or code review of a complex change across multiple reposit", "context": "", "use_case": "planning and complex code review", "switched_from": "", "story": "Why isn't the industry freaking out about DeepSeek 4.1 Flash?"}, {"id": 800299666, "item_id": 50018729, "url": "https://news.ycombinator.com/item?id=50018729", "by": "dada216", "date": "2026-10-09", "stance": "recommend", "quote": "I too find that Opus5.5/Astra/6 are better for UI/UX now", "excerpt": "I even find that for what amounts to a web search Gemini 3.8 flash is better than the big models, stuff like the links to updated tax codes in different countries, or whatever MS is doing with Azure/o365 or AWS with their plethora of products. I think use cases are the real reason why people have such different experiences, I too find that Opus5.5/Astra/6 are better for UI/UX now, it wasn't the case a year ago, at so", "context": "", "use_case": "UI/UX work", "switched_from": "", "story": "Why isn't the industry freaking out about DeepSeek 4.1 Flash?"}, {"id": 800204067, "item_id": 50012754, "url": "https://news.ycombinator.com/item?id=50012754", "by": "gregwebs", "date": "2026-10-08", "stance": "neutral", "quote": "Previously I planned with Opus/Sol/Astra", "excerpt": "I have been using DeepSeek 4.1 flash intensively for over a month. If I run it all day long it costs $1-2. Its fast. Previously I was always quickly running up to my Claude/Codex 5 hour window (on the $20/month plan). The cost savings of DeepSeek is real as shown in this article and I am using subsidized plans. DeepSeek is horrible at grilling sessions (the /grill* skills to make technical decisions). It doesn't know", "context": "", "use_case": "planning and review", "switched_from": "", "story": "Why isn't the industry freaking out about DeepSeek 4.1 Flash?"}, {"id": 800055937, "item_id": 50003496, "url": "https://news.ycombinator.com/item?id=50003496", "by": "flossly", "date": "2026-10-08", "stance": "warn", "quote": "never got past like 84% compat.", "excerpt": "I found the whole section super interesting... I'll copy it here: I used a lot of OpenAI models to try and complete this port. In total I did over $400,000 in API priced tokens with GPT-5.6 Sol and GPT 6 Astra. They wrote over 1.3m lines of Rust over multiple months of /goal loops and never got past like 84% compat. When I saw how little my Claude Code limits were burning, I figured it'd be fun to throw Opus 5.5 at t", "context": "", "use_case": "porting TypeScript compiler to Rust", "switched_from": "", "story": "Port of the TypeScript compiler, checker and lsp to Rust, by LLM"}, {"id": 800031585, "item_id": 50001974, "url": "https://news.ycombinator.com/item?id=50001974", "by": "snlr", "date": "2026-10-08", "stance": "neutral", "quote": "The code itself is written almost exclusively by Astra 6.", "excerpt": "\"Publisher\" was created as a plugin for ChatGPT's web UI. It's a regular MCP server, so any client is technically supported. The basic thought is making website publishing \"zero-click\" by sending the data to the user's own Cloudflare account directly. I'm assuming that I introduced specific security issues while simplifying this workflow. Docs note that users currently have to place considerable trust in the vendor. ", "context": "", "use_case": "writing code", "switched_from": "", "story": "Built a ChatGPT-to-Cloudflare publisher, looking for supplychain/security input"}, {"id": 800024690, "item_id": 50001543, "url": "https://news.ycombinator.com/item?id=50001543", "by": "bitexploder", "date": "2026-10-08", "stance": "recommend", "quote": "Astra is current a bit better than Opus 5.5.", "excerpt": "I have been using these models to make 3d models with build123d \u2014 they are great at it. I am never touching Fusion again. Astra is current a bit better than Opus 5.5.", "context": "", "use_case": "generating 3D models", "switched_from": "", "story": "Claude Haiku 5.5"}, {"id": 799543121, "item_id": 49971445, "url": "https://news.ycombinator.com/item?id=49971445", "by": "jychang", "date": "2026-10-05", "stance": "recommend", "quote": "I used OpenAI Dots running GPT-6 Astra on the Ultra reasoning mode", "excerpt": "I used OpenAI Dots running GPT-6 Astra on the Ultra reasoning mode, as an agent swarm with 6 subagents, to create a new mathematical proof. The proof comes with Lean verification. Caveat: this is not peer reviewed yet, so consider it speculative. Still, not bad for a weekend of playing with agent orchestration. I am in touch with a few mathematicians from the Covering Repository, who have confirmed that the result is", "context": "", "use_case": "math proofs", "switched_from": "", "story": "Using OpenAI Dots as an agent swarm to prove new math, for free (Lean verified)"}, {"id": 799427747, "item_id": 49964234, "url": "https://news.ycombinator.com/item?id=49964234", "by": "Livia_Zaharia", "date": "2026-10-05", "stance": "neutral", "quote": "Initial motivation were success stories of GPT Astra in creating 3d worlds so we decided to give it a try.", "excerpt": "Hello, I am Livia Zaharia, officially an architect, also a parametric jewelery designer, machine learning enthusiast and a startup founder. What I wanted to share is the link above, where, to promote my pieces in a more interactive way I decided to try building from scratch a 3d version of a town where some pieces become buildings. I developed it together with my boyfriend being bord from waiting for our ML training ", "context": "architect, jewelry designer, startup founder", "use_case": "creating 3D worlds", "switched_from": "", "story": null}, {"id": 799009539, "item_id": 49938096, "url": "https://news.ycombinator.com/item?id=49938096", "by": "roosterIllusi0n", "date": "2026-10-02", "stance": "compare", "quote": "I can compare what I use at home vs paid models like astra at work and the difference is mostly meaningless.", "excerpt": "I am using qwen3.8-27B-UD-Q3_K_XL on a 5080 16gb card with 16gb of system ram at around 43-45tok/s with 65k context. I am using that model to set up containers on a proxmox server with two b70s and 96gb ram. The Q3 model implemented 10 different chat models, 2 image models, and 7 web based harnesses so I can compare. It can rewrite any part to improve it. qwen3.8 is more than capable for software dev. The gemma model", "context": "", "use_case": "work coding", "switched_from": "", "story": "Several vulnerabilities have been discovered in the Linux kernel"}, {"id": 798881461, "item_id": 49930091, "url": "https://news.ycombinator.com/item?id=49930091", "by": "nl", "date": "2026-10-02", "stance": "compare", "quote": "Astra can solve problems that Sol can't but for some reason on easy stuff makes uglier solutions.", "excerpt": "Having used them heavily for 3D model understanding since February I can say it's nuanced. Opus 4.6 and 4.7 were bad, but GPT 5.2 and above were very usable. Opus 4.8 was usable, but the GPT 5.x series was better. Fable is great. Opus 5.0 was interesting. It could solve some problems that Sol 5.x couldn't solve (applying a G2 curve on a 3 way corner where one face was a Bezier curve) but you had to be super prescript", "context": "", "use_case": "3D model design", "switched_from": "", "story": "Using Opus 5.5 to discover a new eyewitness record of the dodo"}, {"id": 798623507, "item_id": 49913969, "url": "https://news.ycombinator.com/item?id=49913969", "by": "drusepth", "date": "2026-09-30", "stance": "neutral", "quote": "while I testdrive Astra", "excerpt": "What basic features are missing from agy? I've been using it and cli-cc + web-cc for months (among a few other random harnesses to test here and there) and they all seem roughly comparable to me. I actually just cancelled Ultra also because I couldn't subscribe to a YouTube Family plan while I had it active (Google... :[) but trying to use Codex as a replacement while I testdrive Astra makes me yearn for agy again.", "context": "", "use_case": "", "switched_from": "", "story": "Gemini 4 Argon"}, {"id": 798602385, "item_id": 49912649, "url": "https://news.ycombinator.com/item?id=49912649", "by": "seunosewa", "date": "2026-09-30", "stance": "neutral", "quote": "If a commit was written with Fable/Opus, I use Sol/Astra to write a well reasoned commit message.", "excerpt": "I use a different LLM family to review commits and write detailed descriptions. If a commit was written with Fable/Opus, I use Sol/Astra to write a well reasoned commit message. If the message doesn't match my intent, then that triggers a manual review.", "context": "", "use_case": "reviewing commits and writing messages", "switched_from": "", "story": "Commit description as a thinking tool"}, {"id": 798599520, "item_id": 49912470, "url": "https://news.ycombinator.com/item?id=49912470", "by": "astro1234", "date": "2026-09-30", "stance": "recommend", "quote": "I used Astra to help me build an immersive VR experience", "excerpt": "I used Astra to help me build an immersive VR experience where you can watch the universe evolve from ~50M years old until the present day. Uses the TNG100-2 simulation. Works in browser too!", "context": "", "use_case": "building VR cosmology experience", "switched_from": "", "story": "Immersive Cosmology"}, {"id": 798569553, "item_id": 49910597, "url": "https://news.ycombinator.com/item?id=49910597", "by": "gertlabs", "date": "2026-09-30", "stance": "recommend", "quote": "Both Astra and 6.1-Sol are pretty comfortably ahead of Opus 5.5", "excerpt": "We tested it across 100 unsaturated coding and engineering environments. Both Astra and 6.1-Sol are pretty comfortably ahead of Opus 5.5 in these types of evaluations, and both end up being cheaper than Opus via API usage. 6.1-Sol is also cheaper than Sonnet 5.5 and much smarter. The only verifiable domain Anthropic seems to be clearly ahead is chemistry (and perhaps also some unverifiable domains like being pleasant", "context": "", "use_case": "coding and engineering tasks", "switched_from": "", "story": "GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price"}, {"id": 798523601, "item_id": 49907725, "url": "https://news.ycombinator.com/item?id=49907725", "by": "exfalso", "date": "2026-09-30", "stance": "recommend", "quote": "I use Astra xhigh, sometimes max, never ran out of tokens on the 100$ thing.", "excerpt": "what. I use Astra xhigh, sometimes max, never ran out of tokens on the 100$ thing. I'm using pi though which is by definition harder better faster stronger than claude code/codex.", "context": "", "use_case": "coding", "switched_from": "", "story": "GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price"}, {"id": 798439074, "item_id": 49902442, "url": "https://news.ycombinator.com/item?id=49902442", "by": "srmckee", "date": "2026-09-29", "stance": "recommend", "quote": "so I switched to Astra which is awesome", "excerpt": "Hey HN! Spencer here - wanted to share a token compression tool that I\u2019ve built for myself to save 30% costs on codex! I've been building a ton of web apps to replace saas i'm too cheap to keep paying for like chatgpt, notion, myfitnesspal, etc. and so i can get the exact features i want (+ no ads!). I was using Sol originally which was ok but also irritating because it would mess up and make a buggy ui. so I switche", "context": "", "use_case": "coding model", "switched_from": "Sol", "story": "Show HN: Free token compression CLI, saves codex bills by 30%"}, {"id": 798388800, "item_id": 49899300, "url": "https://news.ycombinator.com/item?id=49899300", "by": "moshegramovsky", "date": "2026-09-29", "stance": "warn", "quote": "I'm not going to keep using Astra unless I can get some benefit from it", "excerpt": "100% hard agree. I used about 10 hours of Astra high-thinking compute time and it was a bad experience. Incredibly slow (prompts running for 30/40 minutes) to do simple things. As a result, Astra didn't get much done. It needs the same small implementation slices as GPT 5.5/others, but was much slower and didn't generate better results. (On a complex infra project/across a large codebase.) It was absolutely terrible ", "context": "pays $200/month for OpenAI", "use_case": "large codebase coding", "switched_from": "", "story": "GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price"}, {"id": 798380226, "item_id": 49898764, "url": "https://news.ycombinator.com/item?id=49898764", "by": "KronisLV", "date": "2026-09-29", "stance": "recommend", "quote": "Astra is a pretty good model", "excerpt": "> the only people I hear talking about actually switching away Opus 5 had really bad writing so I switched to OpenAI, though Opus 5.5 largely addresses that and it's not like them building some Slack integrations (or any other non-core stuff) halts the actual model training in any capacity. For what it's worth, Astra is a pretty good model and for all I know the new Sol will be as well, it's just that it's getting mo", "context": "", "use_case": "", "switched_from": "Opus 5", "story": "Dots: Always-on agents"}, {"id": 798380080, "item_id": 49898755, "url": "https://news.ycombinator.com/item?id=49898755", "by": "johnsmith1840", "date": "2026-09-29", "stance": "neutral", "quote": "then switched to 2 openai for astra", "excerpt": "Yeah I think you're right I have around 500$ a mo budget for AI. Had both a anthropic and openai then switched to 2 openai for astra. They want to capture the straddlers between both. But this is a horrible decision. They could have eased customers into this instead of massive change. The are literally more than doubling the cost for no value add. A smart company would have done this when they released astra or only ", "context": "has around $500 a month AI budget", "use_case": "", "switched_from": "", "story": "ChatGPT Pro 500"}, {"id": 798371552, "item_id": 49898222, "url": "https://news.ycombinator.com/item?id=49898222", "by": "trentnix", "date": "2026-09-29", "stance": "recommend", "quote": "My experience with Astra (I use it at home writing Go and C) for coding has been fantastic.", "excerpt": "That's not been my experience. My experience with Astra (I use it at home writing Go and C) for coding has been fantastic. Opus 5.5 (I use it for work writing C#) seems faster than Opus 5, but it doesn't seem demonstrably better to my eyes and is still prone to word vomit.", "context": "", "use_case": "coding in Go and C", "switched_from": "", "story": "GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price"}, {"id": 798336018, "item_id": 49896001, "url": "https://news.ycombinator.com/item?id=49896001", "by": "dannyw", "date": "2026-09-29", "stance": "neutral", "quote": "Astra being a bit of a \"over-correct over-validation nit-picker\"", "excerpt": "For me, commits are the atomic block; and a good commit should be self-contained anyway; where no additional context is necessary. If an agent can't figure out what a commit is supposed to do, with only the commit title/description and diff, it's a bad commit, and this has always been true in software engineering. I do not pass prompts between the two ordinarily. After claude or codex finishes a commit, I switch cons", "context": "", "use_case": "coding review", "switched_from": "", "story": "Claude partial outage"}, {"id": 798274051, "item_id": 49892128, "url": "https://news.ycombinator.com/item?id=49892128", "by": "joshstrange", "date": "2026-09-29", "stance": "warn", "quote": "tried Astra and it also burned though my weekly usage in a day, with a _single_ agent.", "excerpt": "I can believe that (20x is not 4x of the 5x plan) but I\u2019ve been on the $100/mo Codex and the $200/mo Claude plans and I can run the whole week with multiple agents on Claude, mean while I burned through my whole week in 1 day on Sol. Yes, it\u2019s half the price of my Claude plan but even if OpenAi actually gave 4x usage for the next plan up (which they just halved), it would still be less than the usage I get from Claud", "context": "", "use_case": "coding", "switched_from": "", "story": "OpenAI: Tomorrow we are re-opening the Pro $200 subscription"}, {"id": 798094976, "item_id": 49880936, "url": "https://news.ycombinator.com/item?id=49880936", "by": "andy_ppp", "date": "2026-09-28", "stance": "recommend", "quote": "I asked Astra to come up with a plan for Playwright testing the whole App", "excerpt": "I don't really understand all the negativity - I've delivered amazing robust solutions at about 10x my previous rate I think and can make dramatic cross code changes on awful legacy code bases with ease. I just added some complex prev/next navigation across a load of detail pages pages that rely on the current search, a new PDF generation system that uses a queue and workers and absolutely tones of changes to a legac", "context": "", "use_case": "planning tests and coding", "switched_from": "", "story": "The problem is not AI code, but not knowing about system architecture or intent"}, {"id": 798092532, "item_id": 49880783, "url": "https://news.ycombinator.com/item?id=49880783", "by": "caidehen", "date": "2026-09-28", "stance": "neutral", "quote": "astra designed the project and wrote spec.md", "excerpt": "Agree with the author. I used to write code by hand, literally hand, I used vanilla vim with only syntax highlight and line number enabled, no more. So I know I can write code. I wrote C and Python most. I used to vibe code, I am still vibe coding, multiple projects at same time. I use opus, astra, luna, deepseek v4.1 flash, I use claude code, codex, pi. I vibed a project to manage my coding agents' sub agents, skill", "context": "", "use_case": "project design and specs", "switched_from": "", "story": "Coding is not solved"}, {"id": 797999233, "item_id": 49874952, "url": "https://news.ycombinator.com/item?id=49874952", "by": "sitsitstay", "date": "2026-09-28", "stance": "warn", "quote": "I used Astra, and it lasted less than hour before it was dead for a month", "excerpt": "I've been tokenmaxxing across Claude (multiple subs, but I've now burned all of them - and also exhausted, for all of them, the discounted \"Buy Credits\" from 40% to 25% to 20%....). I used Astra, and it lasted less than hour before it was dead for a month - and Grok whilst did solid work, struggled with cohesive architecture. DeepSeek, well, I've got 400+ dollars left in credit I will donate to my enemies, to slow th", "context": "", "use_case": "coding", "switched_from": "", "story": "Open Annihilation \u2013 Spent the GDP of a Small Nation Remaking a Classic 90s Game"}, {"id": 797994051, "item_id": 49874628, "url": "https://news.ycombinator.com/item?id=49874628", "by": "Gareth321", "date": "2026-09-28", "stance": "warn", "quote": "Even Astra has been noticeably degraded for me and everyone else I have asked.", "excerpt": "They're impossible to objectively track by design and intent. Intelligence is such a nebulous target that one can find any number of metrics to support any premise: that the model is smarter, and that the model is dumber. We use benchmarks to attempt to standardise comparisons but these are quickly ingested into the training data and then become effectively useless. You might have heard the term \"bench-maxed.\" Meanin", "context": "", "use_case": "", "switched_from": "", "story": "Ember-1"}, {"id": 797576497, "item_id": 49848531, "url": "https://news.ycombinator.com/item?id=49848531", "by": "atemerev", "date": "2026-09-25", "stance": "recommend", "quote": "I used Fable 5.1 and Astra to solve and prove a long-standing mathematical problem", "excerpt": "Just recently I used Fable 5.1 and Astra to solve and prove a long-standing mathematical problem I was always interested in: the shoreline search problem (a ship in total fog is at unknown distance from the shore [infinite line], what is the optimal trajectory?) A particular kind of logarithmic spiral was conjectured 33 years ago; a few days ago, I have obtained the proof. How routine it has become. https://arxiv.org", "context": "mathematician", "use_case": "mathematical proofs", "switched_from": "", "story": "Yes, Claude can do nine loops"}, {"id": 796950400, "item_id": 49809400, "url": "https://news.ycombinator.com/item?id=49809400", "by": "amluto", "date": "2026-09-22", "stance": "warn", "quote": "I use Astra for rapidly consuming my token limit on a task that would not consume it on 5.6 Sol.", "excerpt": "I use Astra for rapidly consuming my token limit on a task that would not consume it on 5.6 Sol. (I have not done anything quantitative here. For one thing, OpenAI\u2019s billing pages and the codex-rs frontend make it pathetically difficult to get any real data. Some day I should wire up a proxy to extract actual stats.)", "context": "", "use_case": "", "switched_from": "", "story": "GPT-6 Sol and Luna"}, {"id": 796907472, "item_id": 49806717, "url": "https://news.ycombinator.com/item?id=49806717", "by": "Rapzid", "date": "2026-09-22", "stance": "recommend", "quote": "I use Astra for destroying vaguely scoped asks and tasks, and then for high-level design and plan generations", "excerpt": "Yeah, I use Astra for destroying vaguely scoped asks and tasks, and then for high-level design and plan generations.. Otherwise I'm using 5.6 Sol for actual plan execution and review..", "context": "", "use_case": "planning and design", "switched_from": "", "story": "GPT-6 Sol and Luna"}, {"id": 796896819, "item_id": 49806051, "url": "https://news.ycombinator.com/item?id=49806051", "by": "hombre_fatal", "date": "2026-09-22", "stance": "warn", "quote": "Astra blows through usage limits too fast to be a daily driver and writes weird code", "excerpt": "I mainly use Codex/Sol to review my plans drafted by Fable. But beyond that, Astra blows through usage limits too fast to be a daily driver and writes weird code despite what my \"house style\" is, and Codex is behind Claude Code in terms of critical features like seeing what's going on in subagents. The parent + subagent workflow has become critical for keeping the reasoning agent (parent) context-lean while also lett", "context": "", "use_case": "", "switched_from": "", "story": "GPT-6 Sol and Luna"}, {"id": 796660112, "item_id": 49791257, "url": "https://news.ycombinator.com/item?id=49791257", "by": "brink", "date": "2026-09-21", "stance": "warn", "quote": "Astra is extremely RL fried, and nobody is talking about it.", "excerpt": "Same, Astra is extremely RL fried, and nobody is talking about it. I used Astra for a few days on my personal project, and load times went from less than 3 seconds to almost 30 seconds because it kept using the wrong sync primitives and bad architecture overall.", "context": "", "use_case": "personal project coding", "switched_from": "", "story": "Grok 4.7"}, {"id": 796647733, "item_id": 49790483, "url": "https://news.ycombinator.com/item?id=49790483", "by": "swalsh", "date": "2026-09-21", "stance": "warn", "quote": "astra does tend to chew up usage", "excerpt": "Codex has become my goto tooling. I used to be a Claude Max subscriber, but I was becoming disappointed with the quality of the output from Opus 5. Fable chewed through my usage too quickly to be practical. Moving to a Pro account w/ Codex was a big improvement. Sol had great output, and the usage was more than sufficient for most of my needs. However astra does tend to chew up usage, so when i've done to much of tha", "context": "", "use_case": "", "switched_from": "", "story": "Grok 4.7"}, {"id": 795549792, "item_id": 49721862, "url": "https://news.ycombinator.com/item?id=49721862", "by": "pod_krad", "date": "2026-09-16", "stance": "recommend", "quote": "Astra xhigh - for specs, analysis.", "excerpt": "Astra xhigh - for specs, analysis. Sol xhigh - for orchestration and implementation in complicated cases. Terra, Luna - for implementation in simple routine cases.", "context": "", "use_case": "specs and analysis", "switched_from": "", "story": "Ask HN: What default model do you use and why?"}, {"id": 795196928, "item_id": 49699808, "url": "https://news.ycombinator.com/item?id=49699808", "by": "varispeed", "date": "2026-09-14", "stance": "warn", "quote": "With Astra and Fable I am no longer able to finish tasks I used to be able to do and I revert to older models.", "excerpt": "The AI has gotten substantially worse. With Astra and Fable I am no longer able to finish tasks I used to be able to do and I revert to older models. Latest models go into reasoning loops, commit logic errors routinely and cannot even spot them by themselves and many more. For instance I have been trying to come up with generalised model of certain non-linear behaviour and had Astra working on it for 4 days. It produ", "context": "", "use_case": "modeling non-linear behaviour", "switched_from": "", "story": "Fable 5.1 Solves the Cyphral Distich, a 370-year-old cipher"}]}