{"slug": "claude-opus", "name": "Claude Opus", "category": "ai", "category_name": "AI models and assistants", "description": "Anthropic's flagship large language model, used for coding, writing, and agentic tasks.", "recommenders": 376, "commenters": 717, "warners": 117, "score": 273.15, "rising": 92.67, "first": "2025-11-01", "latest": "2026-10-31", "why": "Praised for strong coding results on large projects, clear and reliable writing, and orchestrating swarms of cheaper models.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 388, "neutral": 210, "compare": 78, "warn": 128}, "spark": [7, 16, 26, 24, 21, 22, 8, 57, 51, 30, 56, 70], "listed": true, "loved": ["Agent orchestration", "Ambitious coding"], "caveat": "Condescending tone", "short_quote": "This is the first model to which I send my collection of nearly 900 poems and an extremely simple prompt (in Portuguese), and it manages to produce an impeccable analysis", "icon": true, "domain": "claude.ai", "github_repo": null, "official_url": "https://www.anthropic.com/claude/opus", "pricing": "paid", "aliases": ["Claude Opus", "Opus", "Opus 5", "Opus 5.5", "Opus 4.8", "Claude Opus 4.6", "Claude Opus 4.5", "Claude Opus 5", "Claude Opus 4.8", "Opus 4.6", "Claude Opus 5.5", "Claude Opus 4.7", "Opus 4.5", "Opus 4.7", "Opus 4", "Opus-5"], "verdict": [{"text": "Praised for strong coding results on large projects,", "cites": [{"id": 803655056, "item_id": 50228441, "url": "https://news.ycombinator.com/item?id=50228441", "by": "codazoda", "date": "2026-10-25", "stance": "recommend", "quote": "I used Opus 5.5 on a $20 plan and I didn\u2019t even hit my limits.", "excerpt": "I\u2019ve built a retro operating system and programming environment for ESP32 microcontrollers called BASIC32. I took it to 39% complete, as a language, over 8-weeks. Then, I spent just 50 minutes describing all the missing parts and it finished the language in 1-shot. I didn\u2019t think it would work. It did. I used Opus 5.5 on a $20 plan and I didn\u2019t even hit my limits.", "context": "", "use_case": "building a language/OS project", "switched_from": "", "story": "We should be more ambitious with the software we build"}, {"id": 803646849, "item_id": 50227928, "url": "https://news.ycombinator.com/item?id=50227928", "by": "genxy", "date": "2026-10-25", "stance": "recommend", "quote": "I used Fable 5.1 and Opus 5.5 to build a Python interpreter in Rust. Appears to pass all the 3.12 code I throw at it.", "excerpt": "I used Fable 5.1 and Opus 5.5 to build a Python interpreter in Rust. Appears to pass all the 3.12 code I throw at it. To get a toy version that can run most things takes under 2 hrs. Half of that time is collecting the tests you want to pass, how it should evaluate the code, what your criteria are, etc.", "context": "", "use_case": "building a Python interpreter", "switched_from": "", "story": "We should be more ambitious with the software we build"}, {"id": 801042912, "item_id": 50065182, "url": "https://news.ycombinator.com/item?id=50065182", "by": "9cb14c1ec0", "date": "2026-10-13", "stance": "recommend", "quote": "Opus 5.5 has built Lineage ROMs for me for 3 different devices in the last couple of weeks", "excerpt": "Opus 5.5 has built Lineage ROMs for me for 3 different devices in the last couple of weeks, that do not have official Lineage support or any one else working on them. https://github.com/9cb14c1ec0/android_device_lenovo_sycamore_row_5G-lineage https://github.com/9cb14c1ec0/android_device_tcl_selene_vzw https://github.com/9cb14c1ec0/android_device_lenovo_clove_row_wifi-lineage Quite impressive what these models are cap", "context": "", "use_case": "building Lineage ROMs", "switched_from": "", "story": "I turned my unused Kindle into a photo frame with GPT-6 Astra"}]}, {"text": " clear and reliable writing,", "cites": [{"id": 804471040, "item_id": 50279440, "url": "https://news.ycombinator.com/item?id=50279440", "by": "tern", "date": "2026-10-28", "stance": "recommend", "quote": "I find Opus 5.5 writes and expresses itself more clearly and is reliable for anything I throw at it", "excerpt": "I only use Fable when \"knowing about more stuff\" is important (like research, exploring ideas), but I find Opus 5.5 writes and expresses itself more clearly and is reliable for anything I throw at it, so I use it much more. It's possible Fable is better, but since it word salads more quickly, it's hard to tell.", "context": "", "use_case": "writing and general tasks", "switched_from": "", "story": "Gemini 4 Carbon"}, {"id": 802049024, "item_id": 50128064, "url": "https://news.ycombinator.com/item?id=50128064", "by": "nl", "date": "2026-10-17", "stance": "recommend", "quote": "whereas Opus 5.5 has fixed that", "excerpt": "I agree that \"frontier model\" really does mean \"pushing the frontier of intelligence\", so Fable or Astra. They might mean the intelligence/dollar frontier aka the Pareto frontier, but honestly much of the time that's not a great measure. If you are doing repeated work on the same task (eg call center triage) then you can always beat a large LLM on price by fine tuning a small one. If you are doing knowledge work then", "context": "", "use_case": "knowledge work", "switched_from": "", "story": "Ultramarine 1\u00a0Flash"}]}, {"text": " and orchestrating swarms of cheaper models.", "cites": [{"id": 802679057, "item_id": 50167441, "url": "https://news.ycombinator.com/item?id=50167441", "by": "A_D_E_P_T", "date": "2026-10-20", "stance": "recommend", "quote": "use Opus 5.5 for review/orchestration", "excerpt": "The Google models are great \"general purpose\" and research LLMs, but they're relatively poor at coding. The way I see it, you have two options. 1) Get on the Claude Max 20x plan and use Opus 5.5 for review/orchestration and Haiku 5.5 for grunt work. Haiku 5.5 is decent, cheap, and fast. 2) Deepseek 4.1 Flash in OpenCode. You should also test MiMo v2.6. It's a little bit uneven, but might work well for whatever you're", "context": "", "use_case": "review and orchestration", "switched_from": "", "story": "Ask HN: Best value/money AI coding plan bundle?"}, {"id": 804179968, "item_id": 50261248, "url": "https://news.ycombinator.com/item?id=50261248", "by": "Roark66", "date": "2026-10-27", "stance": "recommend", "quote": "a supervisor model like Opus 5.5 (or fable before they lobotomised it) limited to 200k of context and told to keep it's progress updated in a text file is smart enough", "excerpt": "Actually more AI agents do the work faster and cheaper. I use such orchestration for those 2 reasons. One, a supervisor model like Opus 5.5 (or fable before they lobotomised it) limited to 200k of context and told to keep it's progress updated in a text file is smart enough to understand the whole task. It farms smaller sub tasks to a fleet of 7 Qwen3.8-Flash-Next agents. Not only it uses up 10% of the tokens opus wo", "context": "", "use_case": "supervisor agent orchestration", "switched_from": "", "story": "The Mythical Man-Month in the Age of AI"}, {"id": 802107808, "item_id": 50131738, "url": "https://news.ycombinator.com/item?id=50131738", "by": "aizk", "date": "2026-10-17", "stance": "recommend", "quote": "I use Opus 5.5 and swarms of about 50 deepseek models to reverse engineer old videos games.", "excerpt": "I use Opus 5.5 and swarms of about 50 deepseek models to reverse engineer old videos games. The scale is roughly 600k loc generated from pure asm, simply by knowing what the old compiler is. Truly staggering stuff.", "context": "", "use_case": "reverse engineering old games", "switched_from": "", "story": "DeepSeek-v4.1-Flash-1017"}]}, {"text": " The most common complaints are a condescending tone", "cites": [{"id": 801250880, "item_id": 50078180, "url": "https://news.ycombinator.com/item?id=50078180", "by": "mattlondon", "date": "2026-10-13", "stance": "warn", "quote": "I do not like the way Opus 5.5 communicates with me.", "excerpt": "Agree I do not like the way Opus 5.5 communicates with me. It feels like it is trying to act superior in a condescending way. DO NOT LIKE. I have not used Argon yet but Gemini 3.8 Flash has been a joy to use in comparison (and so fast!) as a daily driver.", "context": "", "use_case": "", "switched_from": "", "story": "Gemini 4 Argon"}, {"id": 801003825, "item_id": 50062739, "url": "https://news.ycombinator.com/item?id=50062739", "by": "mattlondon", "date": "2026-10-12", "stance": "warn", "quote": "I really dislike the way it talks to me (it feels quite condescending, holier-than-thou, uptight etc)", "excerpt": "Gemini 4 Argon release a few days ago was frontier a few days ago. Things move fast. I think some of the benchmarks that Meta didn't pick show Gemini 4 Argon ahead but really the benchmarks are not that great any more and getting increasingly disconnected from reality as people search for \"headroom\" in the evals to avoid saturation I think. Opus 5.5 is \"top\" for example but I really dislike the way it talks to me (it", "context": "", "use_case": "", "switched_from": "", "story": "Muse Wonder"}]}, {"text": " and rudimentary mistakes or weak architectural judgment.", "cites": [{"id": 801026064, "item_id": 50064129, "url": "https://news.ycombinator.com/item?id=50064129", "by": "prmph", "date": "2026-10-12", "stance": "warn", "quote": "Claude Opus 5.5 still makes rudimentary mistakes all the time as I use it.", "excerpt": "I don't know. Claude Opus 5.5 still makes rudimentary mistakes all the time as I use it. More than that, it cannot go beyond a certain level. Like a dog that can only do dog-level actions. I've lost count of the number of times I step back, take a critical look at what it has been hacking away at for hours, which would seem OK, but can be replaced by a much more effective architecture. For the foreseeable future, a h", "context": "", "use_case": "coding", "switched_from": "", "story": "AI is a tool and using it well takes skill"}, {"id": 802515808, "item_id": 50157238, "url": "https://news.ycombinator.com/item?id=50157238", "by": "hollowturtle", "date": "2026-10-19", "stance": "warn", "quote": "Claude Opus sucks at architecture", "excerpt": "> We've vibecoded scripts recently that genuinely are better than what I would have written and I don't think I'm terrible. I'm ok with vibecoding scripts, I love programming the machine and making the end user happy and not dealing with bash! I don't judge scripts quality as long as they are tools in the workflow, I dont care about that, unless they will somehow be part of the production. The name Script implies tha", "context": "", "use_case": "vibecoding scripts", "switched_from": "", "story": "It's still ok to enjoy programming"}]}], "praise": [{"label": "Agent orchestration", "summary": "Commenters use it as the planner, reviewer or supervisor directing cheaper models.", "quotes": [{"id": 802166560, "item_id": 50135410, "url": "https://news.ycombinator.com/item?id=50135410", "by": "aizk", "date": "2026-10-17", "stance": "recommend", "quote": "I use Opus 5.5 as the orchestrator", "excerpt": "Caveat: I get a budget of free ai tokens at work, as I work for an open source AI inference provider. My goal is to reverse engineer an entire gamecube game -- I'm pushing an existing community decomp that was stalled to the finish line, solo, with lots of AI help. This work was directly inspired by https://decomp.dev/ -- which has just shut down a few days ago because of, ironically... people like me who enjoy decom", "context": "works for an open source AI inference provider", "use_case": "orchestrating decompilation agents", "switched_from": "", "story": "DeepSeek-v4.1-Flash-1017"}, {"id": 804179968, "item_id": 50261248, "url": "https://news.ycombinator.com/item?id=50261248", "by": "Roark66", "date": "2026-10-27", "stance": "recommend", "quote": "a supervisor model like Opus 5.5 (or fable before they lobotomised it) limited to 200k of context and told to keep it's progress updated in a text file is smart enough", "excerpt": "Actually more AI agents do the work faster and cheaper. I use such orchestration for those 2 reasons. One, a supervisor model like Opus 5.5 (or fable before they lobotomised it) limited to 200k of context and told to keep it's progress updated in a text file is smart enough to understand the whole task. It farms smaller sub tasks to a fleet of 7 Qwen3.8-Flash-Next agents. Not only it uses up 10% of the tokens opus wo", "context": "", "use_case": "supervisor agent orchestration", "switched_from": "", "story": "The Mythical Man-Month in the Age of AI"}, {"id": 802068864, "item_id": 50129304, "url": "https://news.ycombinator.com/item?id=50129304", "by": "egeozcan", "date": "2026-10-17", "stance": "recommend", "quote": "I use Opus for planning, through API on medium thinking, it doesn't cost anything significant", "excerpt": "I use Opus for planning, through API on medium thinking, it doesn't cost anything significant and the rest can be happily done with deepseek or glm flash. Using sol 6.1 for review loops at the end. All running on pi. A very complex task tends to cost like 10$ max. It's comparable to a subscription with enormous flexibility. It's a steal.", "context": "", "use_case": "planning", "switched_from": "", "story": "DeepSeek-v4.1-Flash-1017"}]}, {"label": "Ambitious coding", "summary": "Commenters report it finishing large builds such as a language, an interpreter and custom ROMs.", "quotes": [{"id": 803655056, "item_id": 50228441, "url": "https://news.ycombinator.com/item?id=50228441", "by": "codazoda", "date": "2026-10-25", "stance": "recommend", "quote": "I used Opus 5.5 on a $20 plan and I didn\u2019t even hit my limits.", "excerpt": "I\u2019ve built a retro operating system and programming environment for ESP32 microcontrollers called BASIC32. I took it to 39% complete, as a language, over 8-weeks. Then, I spent just 50 minutes describing all the missing parts and it finished the language in 1-shot. I didn\u2019t think it would work. It did. I used Opus 5.5 on a $20 plan and I didn\u2019t even hit my limits.", "context": "", "use_case": "building a language/OS project", "switched_from": "", "story": "We should be more ambitious with the software we build"}, {"id": 803646849, "item_id": 50227928, "url": "https://news.ycombinator.com/item?id=50227928", "by": "genxy", "date": "2026-10-25", "stance": "recommend", "quote": "I used Fable 5.1 and Opus 5.5 to build a Python interpreter in Rust. Appears to pass all the 3.12 code I throw at it.", "excerpt": "I used Fable 5.1 and Opus 5.5 to build a Python interpreter in Rust. Appears to pass all the 3.12 code I throw at it. To get a toy version that can run most things takes under 2 hrs. Half of that time is collecting the tests you want to pass, how it should evaluate the code, what your criteria are, etc.", "context": "", "use_case": "building a Python interpreter", "switched_from": "", "story": "We should be more ambitious with the software we build"}, {"id": 801042912, "item_id": 50065182, "url": "https://news.ycombinator.com/item?id=50065182", "by": "9cb14c1ec0", "date": "2026-10-13", "stance": "recommend", "quote": "Opus 5.5 has built Lineage ROMs for me for 3 different devices in the last couple of weeks", "excerpt": "Opus 5.5 has built Lineage ROMs for me for 3 different devices in the last couple of weeks, that do not have official Lineage support or any one else working on them. https://github.com/9cb14c1ec0/android_device_lenovo_sycamore_row_5G-lineage https://github.com/9cb14c1ec0/android_device_tcl_selene_vzw https://github.com/9cb14c1ec0/android_device_lenovo_clove_row_wifi-lineage Quite impressive what these models are cap", "context": "", "use_case": "building Lineage ROMs", "switched_from": "", "story": "I turned my unused Kindle into a photo frame with GPT-6 Astra"}]}, {"label": "Top output quality", "summary": "Several rank it first and say other models seem dumber by comparison.", "quotes": [{"id": 803969232, "item_id": 50248077, "url": "https://news.ycombinator.com/item?id=50248077", "by": "bigwheels", "date": "2026-10-26", "stance": "recommend", "quote": "all other models seems significantly dumber and waste more of my time than Opus 5.5.", "excerpt": "I used to sing the same tune as you are now- \"models are interchangeable\". For today, all other models seems significantly dumber and waste more of my time than Opus 5.5. tl;dr: You should try Opus 5.5 and report back.", "context": "", "use_case": "software engineering", "switched_from": "", "story": "Anthropic S-1"}, {"id": 803494433, "item_id": 50218402, "url": "https://news.ycombinator.com/item?id=50218402", "by": "sail0rm00n", "date": "2026-10-24", "stance": "recommend", "quote": "1. Opus 5.5", "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": 803434784, "item_id": 50214674, "url": "https://news.ycombinator.com/item?id=50214674", "by": "lnenad", "date": "2026-10-23", "stance": "recommend", "quote": "the quality of the output, in most scenarios that I've been exposed to,  has been the best", "excerpt": "It's not just 5.5, every iteration of the frontier moving - Opus has had some edge. I'm sure it's not 100% in every scenario, I love interacting with Codex models more, but the quality of the output, in most scenarios that I've been exposed to, has been the best.", "context": "", "use_case": "coding", "switched_from": "", "story": "Kimi K3.1 scores 55 on Artificial Analysis, placed on the Pareto line"}]}, {"label": "Code review", "summary": "Teams use it to review code and re-review the issues it finds.", "quotes": [{"id": 804172801, "item_id": 50260800, "url": "https://news.ycombinator.com/item?id=50260800", "by": "Roark66", "date": "2026-10-27", "stance": "recommend", "quote": "runs opus or sonet (depending on complexity) to do a review", "excerpt": "Same here.LLMs are really good at code review providing they are configured correctly and have right docs and prompts. Github copilot review is s great example of the worst possible way this can be misused. It is the worst waste of tokens there is. Instead we use a process we created ourselves that runs opus or sonet (depending on complexity) to do a review (taking into account pr description and comments - copilot i", "context": "team with custom LLM review workflow", "use_case": "PR code review", "switched_from": "GitHub Copilot", "story": "I will no longer upload to NEW"}, {"id": 804098416, "item_id": 50256151, "url": "https://news.ycombinator.com/item?id=50256151", "by": "murkt", "date": "2026-10-27", "stance": "recommend", "quote": "I use Opus as both parent and child.", "excerpt": "When I am making agents do a hard thing, I use Opus as both parent and child. I make the child review its code extensively, and then still make the parent read the work and review once again. It is much faster and cheaper than just rolling with no extra reviews, and then in a day I have fifty subtle bugs.", "context": "", "use_case": "agentic coding with reviews", "switched_from": "", "story": "The Mythical Man-Month in the Age of AI"}, {"id": 802679057, "item_id": 50167441, "url": "https://news.ycombinator.com/item?id=50167441", "by": "A_D_E_P_T", "date": "2026-10-20", "stance": "recommend", "quote": "use Opus 5.5 for review/orchestration", "excerpt": "The Google models are great \"general purpose\" and research LLMs, but they're relatively poor at coding. The way I see it, you have two options. 1) Get on the Claude Max 20x plan and use Opus 5.5 for review/orchestration and Haiku 5.5 for grunt work. Haiku 5.5 is decent, cheap, and fast. 2) Deepseek 4.1 Flash in OpenCode. You should also test MiMo v2.6. It's a little bit uneven, but might work well for whatever you're", "context": "", "use_case": "review and orchestration", "switched_from": "", "story": "Ask HN: Best value/money AI coding plan bundle?"}]}, {"label": "Clear writing", "summary": "Commenters say it writes more clearly and that Opus 5.5 fixed Opus 5's poor English.", "quotes": [{"id": 804471040, "item_id": 50279440, "url": "https://news.ycombinator.com/item?id=50279440", "by": "tern", "date": "2026-10-28", "stance": "recommend", "quote": "I find Opus 5.5 writes and expresses itself more clearly and is reliable for anything I throw at it", "excerpt": "I only use Fable when \"knowing about more stuff\" is important (like research, exploring ideas), but I find Opus 5.5 writes and expresses itself more clearly and is reliable for anything I throw at it, so I use it much more. It's possible Fable is better, but since it word salads more quickly, it's hard to tell.", "context": "", "use_case": "writing and general tasks", "switched_from": "", "story": "Gemini 4 Carbon"}, {"id": 802049024, "item_id": 50128064, "url": "https://news.ycombinator.com/item?id=50128064", "by": "nl", "date": "2026-10-17", "stance": "recommend", "quote": "whereas Opus 5.5 has fixed that", "excerpt": "I agree that \"frontier model\" really does mean \"pushing the frontier of intelligence\", so Fable or Astra. They might mean the intelligence/dollar frontier aka the Pareto frontier, but honestly much of the time that's not a great measure. If you are doing repeated work on the same task (eg call center triage) then you can always beat a large LLM on price by fine tuning a small one. If you are doing knowledge work then", "context": "", "use_case": "knowledge work", "switched_from": "", "story": "Ultramarine 1\u00a0Flash"}, {"id": 800780353, "item_id": 50048772, "url": "https://news.ycombinator.com/item?id=50048772", "by": "kozikow", "date": "2026-10-11", "stance": "recommend", "quote": "I don't recall i last few weeks where Opus 5.5 in CC made an obvious lie", "excerpt": "> Almost every time I use an AI chatbot it eventually turns out it's lying to me. I would not trust them with any of the things you mention. If you talk to free chatgpt, sure. But if you launch Opus 5.5 in claude code - for some reason Opus is the smartest for me, even for \"general\" type of questions, when launched from Claude Code. It has harness/prompts optimized for correctness, while more \"customer-grade\" product", "context": "", "use_case": "general questions and coding", "switched_from": "", "story": "The gap in shared understanding of LLM capability is widening"}]}, {"label": "Generous limits", "summary": "Some users rarely hit plan limits and call it token-efficient.", "quotes": [{"id": 804519376, "item_id": 50282461, "url": "https://news.ycombinator.com/item?id=50282461", "by": "llm_nerd", "date": "2026-10-28", "stance": "recommend", "quote": "I use my Claude Max plan endlessly, on enormous problems", "excerpt": "I only use Pro for Google as their models were so far behind, but I tried to give Argon a go, and notably it is only available in the chat interface. One brief conversation, with three exchanges, and my quota was exhausted and it restricted me to Flash-lite for the rest of the day. Rather comical stuff. And it prompts me to upgrade to Ultra for \"5x more usage\". In contrast I use my Claude Max plan endlessly, on enorm", "context": "heavy Claude Max user", "use_case": "large problems", "switched_from": "", "story": "Gemini 4 Carbon"}, {"id": 803655056, "item_id": 50228441, "url": "https://news.ycombinator.com/item?id=50228441", "by": "codazoda", "date": "2026-10-25", "stance": "recommend", "quote": "I used Opus 5.5 on a $20 plan and I didn\u2019t even hit my limits.", "excerpt": "I\u2019ve built a retro operating system and programming environment for ESP32 microcontrollers called BASIC32. I took it to 39% complete, as a language, over 8-weeks. Then, I spent just 50 minutes describing all the missing parts and it finished the language in 1-shot. I didn\u2019t think it would work. It did. I used Opus 5.5 on a $20 plan and I didn\u2019t even hit my limits.", "context": "", "use_case": "building a language/OS project", "switched_from": "", "story": "We should be more ambitious with the software we build"}, {"id": 803797744, "item_id": 50237359, "url": "https://news.ycombinator.com/item?id=50237359", "by": "wycy", "date": "2026-10-26", "stance": "recommend", "quote": "Opus 5.5 fixed all the previous Opus 5 problems, and while Fable still requires credits, there's hardly any need to use it since Opus 5.5 is so good and so token efficient.", "excerpt": "I did leave. Opus was just pumping out so much verbosity and wasting tokens on all that nonsense, plus I was bitter that Anthropic gates the best model behind usage credits, so I switched to ChatGPT just as Astra came out. Then Opus 5.5 came out and I switched back to Ant after 1 month of OAI. Opus 5.5 fixed all the previous Opus 5 problems, and while Fable still requires credits, there's hardly any need to use it si", "context": "", "use_case": "", "switched_from": "ChatGPT", "story": "Listen and read USA nuclear forces Emergency Action Messages, heard on 11175 kHz"}]}], "caveats": [{"label": "Condescending tone", "summary": "Some users dislike how it communicates and say it feels superior or uptight.", "quotes": [{"id": 801250880, "item_id": 50078180, "url": "https://news.ycombinator.com/item?id=50078180", "by": "mattlondon", "date": "2026-10-13", "stance": "warn", "quote": "I do not like the way Opus 5.5 communicates with me.", "excerpt": "Agree I do not like the way Opus 5.5 communicates with me. It feels like it is trying to act superior in a condescending way. DO NOT LIKE. I have not used Argon yet but Gemini 3.8 Flash has been a joy to use in comparison (and so fast!) as a daily driver.", "context": "", "use_case": "", "switched_from": "", "story": "Gemini 4 Argon"}, {"id": 801003825, "item_id": 50062739, "url": "https://news.ycombinator.com/item?id=50062739", "by": "mattlondon", "date": "2026-10-12", "stance": "warn", "quote": "I really dislike the way it talks to me (it feels quite condescending, holier-than-thou, uptight etc)", "excerpt": "Gemini 4 Argon release a few days ago was frontier a few days ago. Things move fast. I think some of the benchmarks that Meta didn't pick show Gemini 4 Argon ahead but really the benchmarks are not that great any more and getting increasingly disconnected from reality as people search for \"headroom\" in the evals to avoid saturation I think. Opus 5.5 is \"top\" for example but I really dislike the way it talks to me (it", "context": "", "use_case": "", "switched_from": "", "story": "Muse Wonder"}]}, {"label": "Mistakes, architecture", "summary": "Users say it still makes rudimentary mistakes and is weak at architecture.", "quotes": [{"id": 801026064, "item_id": 50064129, "url": "https://news.ycombinator.com/item?id=50064129", "by": "prmph", "date": "2026-10-12", "stance": "warn", "quote": "Claude Opus 5.5 still makes rudimentary mistakes all the time as I use it.", "excerpt": "I don't know. Claude Opus 5.5 still makes rudimentary mistakes all the time as I use it. More than that, it cannot go beyond a certain level. Like a dog that can only do dog-level actions. I've lost count of the number of times I step back, take a critical look at what it has been hacking away at for hours, which would seem OK, but can be replaced by a much more effective architecture. For the foreseeable future, a h", "context": "", "use_case": "coding", "switched_from": "", "story": "AI is a tool and using it well takes skill"}, {"id": 802515808, "item_id": 50157238, "url": "https://news.ycombinator.com/item?id=50157238", "by": "hollowturtle", "date": "2026-10-19", "stance": "warn", "quote": "Claude Opus sucks at architecture", "excerpt": "> We've vibecoded scripts recently that genuinely are better than what I would have written and I don't think I'm terrible. I'm ok with vibecoding scripts, I love programming the machine and making the end user happy and not dealing with bash! I don't judge scripts quality as long as they are tools in the workflow, I dont care about that, unless they will somehow be part of the production. The name Script implies tha", "context": "", "use_case": "vibecoding scripts", "switched_from": "", "story": "It's still ok to enjoy programming"}]}, {"label": "Usage limits", "summary": "One user hit a 5-hour limit after about 30 minutes and found it slower.", "quotes": [{"id": 803457889, "item_id": 50216118, "url": "https://news.ycombinator.com/item?id=50216118", "by": "LeFantome", "date": "2026-10-23", "stance": "warn", "quote": "After about 30 minutes I hit a 5 hour limit.", "excerpt": "I have been using DS Flash 4.1 pretty heavily. It runs for hours and hours and is pretty cheap. I realized the other day that I had a $30 a month Claude subscription. So, I fired it up all excited and tried Opus 5.5 on a codebase I was already in with DS. After about 30 minutes I hit a 5 hour limit. The next day I tried again and the same thing happened. And it felt much slower to boot. I was super disappointed.", "context": "", "use_case": "coding", "switched_from": "", "story": "Kimi K3.1 Pro and K3.1 Pro Highspeed"}]}, {"label": "Inconsistent quality", "summary": "One user says it sometimes became noticeably dumber without explanation.", "quotes": [{"id": 804024529, "item_id": 50251533, "url": "https://news.ycombinator.com/item?id=50251533", "by": "lousken", "date": "2026-10-26", "stance": "warn", "quote": "Why is opus 5.5 suddenly a lot dumber? Who knows.", "excerpt": "My issue with anthropic and openai is that if I use a model like e.g. Sonnet 5 I don't always get full Sonnet 5, sometimes it's dumber, sometimes it is smarter (mostly dumber though), while GLM, Kimi and others stay the same. So if I rely on AI for things I better make sure the platform is stable and does not introduce hidden changes. Why is opus 5.5 suddenly a lot dumber? Who knows.", "context": "", "use_case": "", "switched_from": "", "story": "OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?"}]}, {"label": "Often unnecessary", "summary": "Some say cheaper models handle most tasks and Opus is not efficient.", "quotes": [{"id": 803899728, "item_id": 50243733, "url": "https://news.ycombinator.com/item?id=50243733", "by": "Sphax", "date": "2026-10-26", "stance": "recommend", "quote": "Opus 5.5 is more capable", "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": "daily work", "switched_from": "", "story": "Anthropic S-1"}, {"id": 804558305, "item_id": 50284894, "url": "https://news.ycombinator.com/item?id=50284894", "by": "jitl", "date": "2026-10-29", "stance": "recommend", "quote": "it\u2019s not efficient but yeah it\u2019s the best", "excerpt": "i use opus 5.5 to delegate to sol and luna via codex-thread-bridge (https://github.com/saidelike/codex-thread-bridge). these days i\u2019m experimenting with opus using luna as \u201chands\u201d and it\u2019s largely resolved the tokens per output issue with my Anthropic subscription. \u00af\\_(\u30c4)_/\u00af it\u2019s not efficient but yeah it\u2019s the best", "context": "Anthropic subscriber", "use_case": "orchestrating other models", "switched_from": "", "story": "Claude Code regularly sends each background agent\u2019s transcript to its model"}]}, {"label": "Earlier verbosity", "summary": "Opus 5 was criticized for verbosity and wasted tokens before 5.5 fixed it.", "quotes": [{"id": 803797745, "item_id": 50237359, "url": "https://news.ycombinator.com/item?id=50237359", "by": "wycy", "date": "2026-10-26", "stance": "warn", "quote": "Opus was just pumping out so much verbosity and wasting tokens on all that nonsense", "excerpt": "I did leave. Opus was just pumping out so much verbosity and wasting tokens on all that nonsense, plus I was bitter that Anthropic gates the best model behind usage credits, so I switched to ChatGPT just as Astra came out. Then Opus 5.5 came out and I switched back to Ant after 1 month of OAI. Opus 5.5 fixed all the previous Opus 5 problems, and while Fable still requires credits, there's hardly any need to use it si", "context": "", "use_case": "", "switched_from": "", "story": "Listen and read USA nuclear forces Emergency Action Messages, heard on 11175 kHz"}, {"id": 802049025, "item_id": 50128064, "url": "https://news.ycombinator.com/item?id=50128064", "by": "nl", "date": "2026-10-17", "stance": "warn", "quote": "lots of people moved away from Opus 5 because of it's awful English", "excerpt": "I agree that \"frontier model\" really does mean \"pushing the frontier of intelligence\", so Fable or Astra. They might mean the intelligence/dollar frontier aka the Pareto frontier, but honestly much of the time that's not a great measure. If you are doing repeated work on the same task (eg call center triage) then you can always beat a large LLM on price by fine tuning a small one. If you are doing knowledge work then", "context": "", "use_case": "", "switched_from": "", "story": "Ultramarine 1\u00a0Flash"}]}], "who": [{"by": "pawelwentpawel", "context": "programming since age 12", "url": "https://news.ycombinator.com/item?id=50310740", "date": "2026-10-30"}, {"by": "ashdksnndck", "context": "paid user of Gemini and Claude", "url": "https://news.ycombinator.com/item?id=50300487", "date": "2026-10-30"}, {"by": "jitl", "context": "Anthropic subscriber", "url": "https://news.ycombinator.com/item?id=50284894", "date": "2026-10-29"}, {"by": "llm_nerd", "context": "heavy Claude Max user", "url": "https://news.ycombinator.com/item?id=50282461", "date": "2026-10-28"}, {"by": "owkman", "context": "apparently a Google engineer", "url": "https://news.ycombinator.com/item?id=50278529", "date": "2026-10-28"}, {"by": "drakestapleton", "context": "not a coder", "url": "https://news.ycombinator.com/item?id=50263377", "date": "2026-10-27"}, {"by": "Roark66", "context": "team with custom LLM review workflow", "url": "https://news.ycombinator.com/item?id=50260800", "date": "2026-10-27"}, {"by": "boogiefromzk", "context": "built an agentic delegation scheme", "url": "https://news.ycombinator.com/item?id=50251451", "date": "2026-10-26"}, {"by": "Karrot_Kream", "context": "works at a company serving an internal language model", "url": "https://news.ycombinator.com/item?id=50227149", "date": "2026-10-25"}, {"by": "potsandpans", "context": "decompiling a PlayStation game", "url": "https://news.ycombinator.com/item?id=50222286", "date": "2026-10-24"}, {"by": "throw03172019", "context": "works on a Clojure/ClojureScript codebase", "url": "https://news.ycombinator.com/item?id=50208944", "date": "2026-10-23"}, {"by": "XenophileJKO", "context": "30 years of experience", "url": "https://news.ycombinator.com/item?id=50192101", "date": "2026-10-22"}], "tags": ["for orchestrating agents", "for reverse engineering", "for code review", "for UI design", "for CAD scripting"], "use_cases": ["coding", "ai coding", "work coding", "planning", "agentic coding", "daily coding"], "compared": [{"slug": "claude-fable", "name": "Claude Fable", "co": 239, "switched_from": 5, "switched_to": 15}, {"slug": "claude-sonnet", "name": "Claude Sonnet", "co": 135, "switched_from": 9, "switched_to": 6}, {"slug": "claude-code", "name": "Claude Code", "co": 141, "switched_from": 0, "switched_to": 0}, {"slug": "gpt-5", "name": "GPT-5", "co": 104, "switched_from": 0, "switched_to": 3}, {"slug": "codex", "name": "Codex", "co": 86, "switched_from": 2, "switched_to": 6}, {"slug": "gpt-sol", "name": "GPT Sol", "co": 79, "switched_from": 2, "switched_to": 5}, {"slug": "glm", "name": "GLM", "co": 75, "switched_from": 0, "switched_to": 4}, {"slug": "opencode", "name": "OpenCode", "co": 55, "switched_from": 0, "switched_to": 0}], "makers": [{"id": 762631073, "item_id": 47664442, "url": "https://news.ycombinator.com/item?id=47664442", "by": "bcherny", "date": "2026-04-06", "stance": "recommend", "quote": "In this mode, the model decides how long to think for, which tends to work better than fixed thinking budgets across the board.", "excerpt": "Hey all, Boris from the Claude Code team here. I just responded on the issue, and cross-posting here for input. --- Hi, thanks for the detailed analysis. Before I keep going, I wanted to say I appreciate the depth of thinking & care that went into this. There's a lot here, I will try to break it down a bit. These are the two core things happening: > `redact-thinking-2026-02-12` This beta header hides thinking from th", "context": "member of the Claude Code team", "use_case": "adaptive thinking for coding", "switched_from": "", "story": "Claude Code is unusable for complex engineering tasks with the Feb updates"}], "threads": [{"url": "https://news.ycombinator.com/item?id=49672966", "title": "Ask HN: What default model do you use and why?", "mentions": 27}, {"url": "https://news.ycombinator.com/item?id=49803892", "title": "Claude Opus 5.5", "mentions": 22}, {"url": "https://news.ycombinator.com/item?id=49296740", "title": "Why does Opus 5 feel worse to work with?", "mentions": 20}, {"url": "https://news.ycombinator.com/item?id=49525378", "title": "Claude Fable 5.1 and Claude Mythos 5.1", "mentions": 13}, {"url": "https://news.ycombinator.com/item?id=48463808", "title": "Claude Fable 5", "mentions": 13}, {"url": "https://news.ycombinator.com/item?id=47793411", "title": "Claude Opus 4.7", "mentions": 8}], "sentiment_by_month": {"2025-11": {"compare": 1, "recommend": 7}, "2025-12": {"recommend": 16, "neutral": 9}, "2026-01": {"recommend": 26, "neutral": 5, "compare": 1, "warn": 2}, "2026-02": {"neutral": 11, "compare": 4, "recommend": 24, "warn": 3}, "2026-03": {"recommend": 21, "neutral": 9, "compare": 2, "warn": 4}, "2026-04": {"neutral": 12, "warn": 10, "recommend": 22, "compare": 7}, "2026-05": {"warn": 5, "recommend": 8, "neutral": 4, "compare": 7}, "2026-06": {"recommend": 57, "neutral": 36, "compare": 14, "warn": 15}, "2026-07": {"recommend": 51, "warn": 12, "neutral": 35, "compare": 14}, "2026-08": {"compare": 15, "recommend": 30, "neutral": 22, "warn": 40}, "2026-09": {"neutral": 39, "recommend": 56, "warn": 30, "compare": 8}, "2026-10": {"recommend": 70, "neutral": 28, "warn": 7, "compare": 5}}, "quotes": [{"id": 805124433, "item_id": 50320277, "url": "https://news.ycombinator.com/item?id=50320277", "by": "gambiting", "date": "2026-10-31", "stance": "neutral", "quote": "I use Opus at work since we have a nearly unlimited AI budget", "excerpt": "GLM 5.3 Flash is incredibly cheap and imho as good as or better than Sonnet. I used it recently to fix an old server that my family had running serving some websites, and by fix it I mean I got it to log in over ssh, figure out what's running there, copy it out, log in to a brand new droplet on DO, set up a fresh instance of apache with everything patched up to latest, copy that ancient content over, make sure it run", "context": "", "use_case": "work coding", "switched_from": "", "story": "HackerTube: YouTube built from videos recommended by HN"}, {"id": 804998176, "item_id": 50312386, "url": "https://news.ycombinator.com/item?id=50312386", "by": "plaguuuuuu", "date": "2026-10-31", "stance": "neutral", "quote": "The way I use Opus is basically adhering to its training/emergent behavior.", "excerpt": "The way I use Opus is basically adhering to its training/emergent behavior. Anything else is always worse in some way. Using default harness, mostly letting it do what it wants the way it wants to (as much as possible anyway) and making it work for my own flow. For sure it's annoying building up so much knowledge about a single product that changes every 6m anyway, but what can you do.", "context": "", "use_case": "coding", "switched_from": "", "story": "Anthropic to receive high 'junk' rating from credit agencies following IPO"}, {"id": 804993474, "item_id": 50312092, "url": "https://news.ycombinator.com/item?id=50312092", "by": "CharlieDigital", "date": "2026-10-31", "stance": "compare", "quote": "I prefer Anthropic models generally only when the task is more open ended and exploratory.", "excerpt": "I genuinely prefer OpenAI models for coding and building in general. My observation: GPT models are generally better at following the system prompt, tool use, and skill use. I use Paseo and there is a noticeable difference when GPT-6.1 Sol is driving the orchestrator vs Opus/Sonnet 5.5 driving the orchestrator. Anthropic models ignore the shared system prompt to use the Paseo tools for sub-agents and prefer to use it", "context": "", "use_case": "open-ended coding", "switched_from": "", "story": "Anthropic to receive high 'junk' rating from credit agencies following IPO"}, {"id": 804971840, "item_id": 50310740, "url": "https://news.ycombinator.com/item?id=50310740", "by": "pawelwentpawel", "date": "2026-10-30", "stance": "recommend", "quote": "lots of props to Opus 5.5 - it was incredibly good at getting the right game play feel from the start", "excerpt": "Thanks! To respond to your questions - there is a lot indeed. I think the trick was reusing the same techniques I used before (materials, shaders, colours, light setups and much more) from my previous project (https://flat.social - spatial audio and video in 3d, much more complex). I reused a lot of already optimised elements from there and the general look & feel that took me quite a lot of time to arrive at. Also l", "context": "programming since age 12", "use_case": "game development", "switched_from": "", "story": "Show HN: I built a multiplayer game where you're an aerial fighter cat"}, {"id": 804918432, "item_id": 50307402, "url": "https://news.ycombinator.com/item?id=50307402", "by": "dbreunig", "date": "2026-10-30", "stance": "recommend", "quote": "", "excerpt": "A couple thoughts on the process here\u2026 First off, Opus 5.5 built working weights and code with only two prompts: 1. Read this article and any relevant source code: https://divagr.com/blog/qwenputer. Then explore if you could emulate the chip and ram in a Nintendo Entertainment System (as people do today with FPGAs in MISTRs) using this technique. If so, what size model would you have to build and if you could use sma", "context": "", "use_case": "building NES emulator weights", "switched_from": "", "story": "Show HN: A working Nintendo in ~1GB of Qwen weights"}, {"id": 804807792, "item_id": 50300487, "url": "https://news.ycombinator.com/item?id=50300487", "by": "ashdksnndck", "date": "2026-10-30", "stance": "recommend", "quote": "I use opus 5.5 for any question that Google could plausibly screw up.", "excerpt": "If I have a really straightforward question and I want a fast answer I\u2019ll sometimes Google search for the AI overview. But it\u2019s so unreliable. I use opus 5.5 for any question that Google could plausibly screw up. This is a huge product failure on Google\u2019s part, IMO. I\u2019m a $20/paid user for both Gemini and Claude. It\u2019s nuts that with Google\u2019s massive head-start in AI (and apparently they still are capable of making a ", "context": "paid user of Gemini and Claude", "use_case": "answering questions", "switched_from": "", "story": "Google Search is now less than half of Alphabet\u2019s revenue"}, {"id": 804750816, "item_id": 50296926, "url": "https://news.ycombinator.com/item?id=50296926", "by": "jrockway", "date": "2026-10-29", "stance": "recommend", "quote": "Opus 5.5 is definitely good.", "excerpt": "Opus 5.5 is definitely good. I use both all day and it's really hard to say which is which from looking at the outputs; I can't read a PR and say \"this was Fable\" or \"this was Opus\" like I could pre-5.5. Fable is better at playing along with my \"You are a dog\" instruction in CLAUDE.md. It is willing to be fun and playful, whereas Opus just wants to finish the job and go home. Friend vs. coworker, basically. For examp", "context": "", "use_case": "coding", "switched_from": "", "story": "Ask HN: Opus 5.5 suddenly dumb as rocks?"}, {"id": 804739552, "item_id": 50296222, "url": "https://news.ycombinator.com/item?id=50296222", "by": "Oscalemor", "date": "2026-10-29", "stance": "compare", "quote": "ultimately by accountant gave a fairly different view compared to Opus 5 at the time", "excerpt": "I love this type of benchmarking, especially when it to me was counter intuitive that astra missed it. Got a couple of nice prompts and things to think about for my own tax making moving to a new country, but honestly hard for most people to review if they got the optimal output from taxes. Since it's really a proving game, you can optimise and take more risks with declaration, and some of it has to be proven in an a", "context": "moving to a new country", "use_case": "tax advice", "switched_from": "", "story": "Dayjob: Tax \u2013 evaluating AI agents on 50 professional tax assignments"}, {"id": 804705251, "item_id": 50294078, "url": "https://news.ycombinator.com/item?id=50294078", "by": "Revanche1367", "date": "2026-10-29", "stance": "recommend", "quote": "it hallucinates significantly more even on moderately complex tasks compared to Astra or Opus", "excerpt": "I had been on the pro plan for a few months and recently upgraded to the $200.00 AI Ultra plan because I've gotten good mileage out of 3.8 flash while keeping in mind that it's far from perfect. Argon is better in my experience in terms of reliability (especially in not claiming to have done things it clearly didn't) but I have to agree with the hallucination part, for a \"SOTA\" model (before Carbon came out at least)", "context": "", "use_case": "complex tasks", "switched_from": "", "story": "Gemini 4 Carbon is the new #1 on the Artificial Analysis Intelligence Index v5"}, {"id": 804684145, "item_id": 50292759, "url": "https://news.ycombinator.com/item?id=50292759", "by": "oblio", "date": "2026-10-29", "stance": "recommend", "quote": "Opus is much better than it. A lot more features and everything works better", "excerpt": "> Gemini is also IMO excellent for everything that is not coding. It's really rare that I use Anthropic models (which I pay a subscription for) for anything else than coding. I use Gemini Pro for every day stuff and Opus is much better than it. A lot more features and everything works better, especially in the websites. Gemini is just... dumber overall. I had a project/gem (now replaced by a skill) and sometimes I'd ", "context": "", "use_case": "everyday tasks", "switched_from": "Gemini Pro", "story": "Google Search is now less than half of Alphabet\u2019s revenue"}, {"id": 804644785, "item_id": 50290299, "url": "https://news.ycombinator.com/item?id=50290299", "by": "almindor", "date": "2026-10-29", "stance": "recommend", "quote": "I use Opus 5.5 at work and it's still much better than this", "excerpt": "They didn't. Original price was maybe higher but they slashed it during preview and in fact it was lower by a bit ($0.68/M tokens) and $2.09 for output. I've been using it for my personal project and it's still too expensive and very slow when thinking. I think we need one more generation from mistral and with maybe 50% of this already \"slashed\" price for this to be competitive (I use Opus 5.5 at work and it's still ", "context": "", "use_case": "work", "switched_from": "", "story": "We're releasing the weights of Mistral Large 4"}, {"id": 804587153, "item_id": 50286697, "url": "https://news.ycombinator.com/item?id=50286697", "by": "slimbuck", "date": "2026-10-29", "stance": "recommend", "quote": "I just got opus to valid this now and it found another issue with our scale codebook gen", "excerpt": "Hi Wieland and congrats on your paper! The issue you saw had been addressed in 1.7.0. We were generating codebooks for color and scale using k-means, which would result in bad values sometimes for the extremes. In 1.7.0 we switched to an optimal quantiser with outlier handling. It was actually surprisingly hard to get this right. I just got opus to valid this now and it found another issue with our scale codebook gen", "context": "", "use_case": "validating code", "switched_from": "", "story": "Gaussian Splats part 3: There's always a bigger scene"}, {"id": 804561281, "item_id": 50285080, "url": "https://news.ycombinator.com/item?id=50285080", "by": "bitexploder", "date": "2026-10-29", "stance": "recommend", "quote": "Opus 5.5 is a sweet spot as well. I use both extensively.", "excerpt": "I have ended up using Argon more and more. If you keep it grounded it is a great model. Extremely thorough. A little slow. It feels like it gets things done better than Opus 5.5. It slowly churns through everything. Not necessarily quality but its adherence to goals seems better and it takes less prompting. For cyber use cases I am absolutely blown away. Can\u2019t say much more, but it is a serious model. Just my opinion", "context": "", "use_case": "coding", "switched_from": "", "story": "Gemini 4 Carbon"}, {"id": 804558305, "item_id": 50284894, "url": "https://news.ycombinator.com/item?id=50284894", "by": "jitl", "date": "2026-10-29", "stance": "recommend", "quote": "it\u2019s not efficient but yeah it\u2019s the best", "excerpt": "i use opus 5.5 to delegate to sol and luna via codex-thread-bridge (https://github.com/saidelike/codex-thread-bridge). these days i\u2019m experimenting with opus using luna as \u201chands\u201d and it\u2019s largely resolved the tokens per output issue with my Anthropic subscription. \u00af\\_(\u30c4)_/\u00af it\u2019s not efficient but yeah it\u2019s the best", "context": "Anthropic subscriber", "use_case": "orchestrating other models", "switched_from": "", "story": "Claude Code regularly sends each background agent\u2019s transcript to its model"}, {"id": 804519376, "item_id": 50282461, "url": "https://news.ycombinator.com/item?id=50282461", "by": "llm_nerd", "date": "2026-10-28", "stance": "recommend", "quote": "I use my Claude Max plan endlessly, on enormous problems", "excerpt": "I only use Pro for Google as their models were so far behind, but I tried to give Argon a go, and notably it is only available in the chat interface. One brief conversation, with three exchanges, and my quota was exhausted and it restricted me to Flash-lite for the rest of the day. Rather comical stuff. And it prompts me to upgrade to Ultra for \"5x more usage\". In contrast I use my Claude Max plan endlessly, on enorm", "context": "heavy Claude Max user", "use_case": "large problems", "switched_from": "", "story": "Gemini 4 Carbon"}, {"id": 804494593, "item_id": 50280912, "url": "https://news.ycombinator.com/item?id=50280912", "by": "n0on3", "date": "2026-10-28", "stance": "recommend", "quote": "is still vastly capable and I can read its output", "excerpt": "Speaking for a user-facing use, not autonomous coding or agentic workflows: Fable 5.1 choice of words and general \u201cwords salad\u201d expression are horrific. It starts not so good, then as the context grows (still way below any reasonable rotting threshold) it gets completely insane. It\u2019s pointless for a model to be extremely capable if it gives you brain damage. I don\u2019t know if it\u2019s because of the watermark thingy, all I", "context": "", "use_case": "user-facing writing", "switched_from": "", "story": "Gemini 4 Carbon"}, {"id": 804479952, "item_id": 50279997, "url": "https://news.ycombinator.com/item?id=50279997", "by": "RestartKernel", "date": "2026-10-28", "stance": "recommend", "quote": "If the entire industry stopped training new models, and just served Opus 5.5, I would be perfectly satisfied.", "excerpt": "If the entire industry stopped training new models, and just served Opus 5.5, I would be perfectly satisfied. I don't know about its absolute performance, but Opus 5.5 is very \"tasteful\". It came out a week after I switched from Codex and started lamenting how much worse Opus felt to use than Sol, but whatever Anthropic did to fix it worked. Hopefully this isn't a fluke and they'll be able to keep on track with futur", "context": "", "use_case": "AI coding", "switched_from": "Codex", "story": "Opus 5.5 is awesome"}, {"id": 804471040, "item_id": 50279440, "url": "https://news.ycombinator.com/item?id=50279440", "by": "tern", "date": "2026-10-28", "stance": "recommend", "quote": "I find Opus 5.5 writes and expresses itself more clearly and is reliable for anything I throw at it", "excerpt": "I only use Fable when \"knowing about more stuff\" is important (like research, exploring ideas), but I find Opus 5.5 writes and expresses itself more clearly and is reliable for anything I throw at it, so I use it much more. It's possible Fable is better, but since it word salads more quickly, it's hard to tell.", "context": "", "use_case": "writing and general tasks", "switched_from": "", "story": "Gemini 4 Carbon"}, {"id": 804456465, "item_id": 50278529, "url": "https://news.ycombinator.com/item?id=50278529", "by": "owkman", "date": "2026-10-28", "stance": "recommend", "quote": "Opus 5.5 is great, but I use Gemini far more than I do Opus at this point.", "excerpt": "Other comments on this already, so this is more of a confirmation. I'm not going to comment on others use of Carbon except to say some article that came out before Argon's external launch saying Googlers hate it was nonsense. Not to say everyone loved it all the time. I use Carbon heavily and am quite happy with it. For complex designs I may have it and Opus iterate back and forth on the design. They have their own s", "context": "apparently a Google engineer", "use_case": "design iteration", "switched_from": "", "story": "Gemini 4 Carbon"}, {"id": 804454128, "item_id": 50278383, "url": "https://news.ycombinator.com/item?id=50278383", "by": "jassyr", "date": "2026-10-28", "stance": "recommend", "quote": "I use opus 5.5 for legal stuff and it's generally great", "excerpt": "I wonder if you use opus 5.5 and let it research using the internet what you would get. I use opus 5.5 for legal stuff and it's generally great, but you do have to give it access to the web.", "context": "", "use_case": "legal research", "switched_from": "", "story": "Dayjob: Tax \u2013 evaluating AI agents on 50 professional tax assignments"}, {"id": 804444465, "item_id": 50277779, "url": "https://news.ycombinator.com/item?id=50277779", "by": "nater5000", "date": "2026-10-28", "stance": "neutral", "quote": "someone who just sticks with Opus 5.5 mostly across the board", "excerpt": "For what it's worth: Argon is a pretty good model, and this is coming from someone who just sticks with Opus 5.5 mostly across the board. For the current introductory price, it provides an exceptional value. Looks like they're trying to release Carbon (which is apparently a significant improvement) at the same price as Argon, which would be a crazy value. It's worth trying if you have easy access to it (we use Cursor", "context": "", "use_case": "", "switched_from": "", "story": "Gemini 4 Carbon"}, {"id": 804214032, "item_id": 50263377, "url": "https://news.ycombinator.com/item?id=50263377", "by": "drakestapleton", "date": "2026-10-27", "stance": "neutral", "quote": "Opus 5.5 being\u2026 well Opus 5.5 it did it.", "excerpt": "This was not my intention at first. Omega came out of Omegese which was my shorthand I want the agent to use because it was making a lot of errors and throwing my swarm off. Mainly by json{}[]{\\\\]]haf and whatever the hell else it uses. To just a regular engineer, not sure what it means. Just know it was a pain in my ass so I told it to rewrite whatever it wanted to use to stop making errors and Opus 5.5 being\u2026 well ", "context": "not a coder", "use_case": "agent coding", "switched_from": "", "story": "Running Local AI on Nvidia DGX Spark Without CUDA"}, {"id": 804179968, "item_id": 50261248, "url": "https://news.ycombinator.com/item?id=50261248", "by": "Roark66", "date": "2026-10-27", "stance": "recommend", "quote": "a supervisor model like Opus 5.5 (or fable before they lobotomised it) limited to 200k of context and told to keep it's progress updated in a text file is smart enough", "excerpt": "Actually more AI agents do the work faster and cheaper. I use such orchestration for those 2 reasons. One, a supervisor model like Opus 5.5 (or fable before they lobotomised it) limited to 200k of context and told to keep it's progress updated in a text file is smart enough to understand the whole task. It farms smaller sub tasks to a fleet of 7 Qwen3.8-Flash-Next agents. Not only it uses up 10% of the tokens opus wo", "context": "", "use_case": "supervisor agent orchestration", "switched_from": "", "story": "The Mythical Man-Month in the Age of AI"}, {"id": 804172801, "item_id": 50260800, "url": "https://news.ycombinator.com/item?id=50260800", "by": "Roark66", "date": "2026-10-27", "stance": "recommend", "quote": "runs opus or sonet (depending on complexity) to do a review", "excerpt": "Same here.LLMs are really good at code review providing they are configured correctly and have right docs and prompts. Github copilot review is s great example of the worst possible way this can be misused. It is the worst waste of tokens there is. Instead we use a process we created ourselves that runs opus or sonet (depending on complexity) to do a review (taking into account pr description and comments - copilot i", "context": "team with custom LLM review workflow", "use_case": "PR code review", "switched_from": "GitHub Copilot", "story": "I will no longer upload to NEW"}, {"id": 804156465, "item_id": 50259779, "url": "https://news.ycombinator.com/item?id=50259779", "by": "jrmg", "date": "2026-10-27", "stance": "recommend", "quote": "", "excerpt": "I used to think this way too, and I believe myself to have high standards about this kind of thing. But with the very latest models (like, Opus 5.5 class and up), I am now of the opinion that the output, in non-creative situations like code review and most software development, is fine. Great, even. Better than the reviews a lot of humans generate. Honestly, if you\u2019re at all open minded about this, get a $20 Claude P", "context": "", "use_case": "code review and coding", "switched_from": "", "story": "I will no longer upload to NEW"}, {"id": 804098416, "item_id": 50256151, "url": "https://news.ycombinator.com/item?id=50256151", "by": "murkt", "date": "2026-10-27", "stance": "recommend", "quote": "I use Opus as both parent and child.", "excerpt": "When I am making agents do a hard thing, I use Opus as both parent and child. I make the child review its code extensively, and then still make the parent read the work and review once again. It is much faster and cheaper than just rolling with no extra reviews, and then in a day I have fifty subtle bugs.", "context": "", "use_case": "agentic coding with reviews", "switched_from": "", "story": "The Mythical Man-Month in the Age of AI"}, {"id": 804024529, "item_id": 50251533, "url": "https://news.ycombinator.com/item?id=50251533", "by": "lousken", "date": "2026-10-26", "stance": "warn", "quote": "Why is opus 5.5 suddenly a lot dumber? Who knows.", "excerpt": "My issue with anthropic and openai is that if I use a model like e.g. Sonnet 5 I don't always get full Sonnet 5, sometimes it's dumber, sometimes it is smarter (mostly dumber though), while GLM, Kimi and others stay the same. So if I rely on AI for things I better make sure the platform is stable and does not introduce hidden changes. Why is opus 5.5 suddenly a lot dumber? Who knows.", "context": "", "use_case": "", "switched_from": "", "story": "OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?"}, {"id": 804023218, "item_id": 50251451, "url": "https://news.ycombinator.com/item?id=50251451", "by": "boogiefromzk", "date": "2026-10-26", "stance": "neutral", "quote": "I also made sure Opus controls Grok in Cursor", "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": "controlling cheaper model", "switched_from": "", "story": "Cost-effective agentic delegation scheme for Claude Code, Cursor and Codex"}, {"id": 803969232, "item_id": 50248077, "url": "https://news.ycombinator.com/item?id=50248077", "by": "bigwheels", "date": "2026-10-26", "stance": "recommend", "quote": "all other models seems significantly dumber and waste more of my time than Opus 5.5.", "excerpt": "I used to sing the same tune as you are now- \"models are interchangeable\". For today, all other models seems significantly dumber and waste more of my time than Opus 5.5. tl;dr: You should try Opus 5.5 and report back.", "context": "", "use_case": "software engineering", "switched_from": "", "story": "Anthropic S-1"}, {"id": 803920576, "item_id": 50245036, "url": "https://news.ycombinator.com/item?id=50245036", "by": "myphpgothacked", "date": "2026-10-26", "stance": "neutral", "quote": "I spent unbelievable amounts of money on Opus tokens for OpenClaw", "excerpt": "To be fair, I spent unbelievable amounts of money on Opus tokens for OpenClaw. Easily my biggest expense this year. The risk is that other models are pretty comparable to Opus since around July. Unless they have cheaper electricity costs, I can switch models interchangeably. They don't have a moat. Gosh I wish it was easier to short stocks. Whatever platform I used made me sell/buy 100, and I wanted to do this short ", "context": "", "use_case": "powering OpenClaw", "switched_from": "", "story": "Anthropic S-1"}, {"id": 803903025, "item_id": 50243939, "url": "https://news.ycombinator.com/item?id=50243939", "by": "criley2", "date": "2026-10-26", "stance": "neutral", "quote": "I use Opus 5.5 at work and I use Sonnet 5.5 at home.", "excerpt": "I use Opus 5.5 at work and I use Sonnet 5.5 at home. Sonnet 5.5 medium according to AA cost per task is $0.48 per task, while GLM 5.3 (max) is $2.01 (opus 5.5 high is $1.82). GLM 5.3 flash is $0.28 per task and Deepseek 4.1 flash 1017 is $0.26. So it seems like Sonnet 5.5 medium, GLM 5.3 flash and Deepseek V4.1 flash 1017 are all in a similar space in capacity and price, however the $20 anthropic sub will get you lik", "context": "", "use_case": "work", "switched_from": "", "story": "Anthropic S-1"}, {"id": 803899728, "item_id": 50243733, "url": "https://news.ycombinator.com/item?id=50243733", "by": "Sphax", "date": "2026-10-26", "stance": "recommend", "quote": "Opus 5.5 is more capable", "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": "daily work", "switched_from": "", "story": "Anthropic S-1"}, {"id": 803853442, "item_id": 50240840, "url": "https://news.ycombinator.com/item?id=50240840", "by": "mattlondon", "date": "2026-10-26", "stance": "neutral", "quote": "Yes I have tried Opus 5.5", "excerpt": "I think, despite the desperate attempt at IBM to remain relevant (and in business), their CEO has a point here that do you really need SOTA models for everything? I had a really nice time with GLM5.2 for a month or two - it was, in my experience, approx Opus 4.5ish level of capability when I think Opus 4.8 was the new hotness Even today, I use Gemini 3.8 Flash as my daily driver for most coding. Its fast and good, ev", "context": "", "use_case": "coding", "switched_from": "", "story": "IBM CEO says open weight models are now good and cheap on-prem"}, {"id": 803830658, "item_id": 50239416, "url": "https://news.ycombinator.com/item?id=50239416", "by": "ipsum2", "date": "2026-10-26", "stance": "neutral", "quote": "I use Kimi, chatGPT (Astra/Sol), and Claude (Opus 5.5)", "excerpt": "> much cheaper to access I'm calling BS. I use Kimi, chatGPT (Astra/Sol), and Claude (Opus 5.5), and Kimi is by far the most expensive in terms of tokens and $/work done. Using OpenCode harness.", "context": "", "use_case": "coding work", "switched_from": "", "story": "OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?"}, {"id": 803797744, "item_id": 50237359, "url": "https://news.ycombinator.com/item?id=50237359", "by": "wycy", "date": "2026-10-26", "stance": "recommend", "quote": "Opus 5.5 fixed all the previous Opus 5 problems, and while Fable still requires credits, there's hardly any need to use it since Opus 5.5 is so good and so token efficient.", "excerpt": "I did leave. Opus was just pumping out so much verbosity and wasting tokens on all that nonsense, plus I was bitter that Anthropic gates the best model behind usage credits, so I switched to ChatGPT just as Astra came out. Then Opus 5.5 came out and I switched back to Ant after 1 month of OAI. Opus 5.5 fixed all the previous Opus 5 problems, and while Fable still requires credits, there's hardly any need to use it si", "context": "", "use_case": "", "switched_from": "ChatGPT", "story": "Listen and read USA nuclear forces Emergency Action Messages, heard on 11175 kHz"}, {"id": 803797745, "item_id": 50237359, "url": "https://news.ycombinator.com/item?id=50237359", "by": "wycy", "date": "2026-10-26", "stance": "warn", "quote": "Opus was just pumping out so much verbosity and wasting tokens on all that nonsense", "excerpt": "I did leave. Opus was just pumping out so much verbosity and wasting tokens on all that nonsense, plus I was bitter that Anthropic gates the best model behind usage credits, so I switched to ChatGPT just as Astra came out. Then Opus 5.5 came out and I switched back to Ant after 1 month of OAI. Opus 5.5 fixed all the previous Opus 5 problems, and while Fable still requires credits, there's hardly any need to use it si", "context": "", "use_case": "", "switched_from": "", "story": "Listen and read USA nuclear forces Emergency Action Messages, heard on 11175 kHz"}, {"id": 803692337, "item_id": 50230771, "url": "https://news.ycombinator.com/item?id=50230771", "by": "holbrad", "date": "2026-10-25", "stance": "neutral", "quote": "we regularly improve the agents' code and design though human review, even with Opus 5.5.", "excerpt": "I couldn't agree more around the ideological friction slowing things down, especially code review. I'd be really curious to get people's thoughts. Personally I do code review, it's what we do at my workplace, and we regularly improve the agents' code and design though human review, even with Opus 5.5. But looking at some of the people who are pushing AI development the furthest, I do see the trend towards it just not", "context": "", "use_case": "coding agent", "switched_from": "", "story": "We should be more ambitious with the software we build"}, {"id": 803655056, "item_id": 50228441, "url": "https://news.ycombinator.com/item?id=50228441", "by": "codazoda", "date": "2026-10-25", "stance": "recommend", "quote": "I used Opus 5.5 on a $20 plan and I didn\u2019t even hit my limits.", "excerpt": "I\u2019ve built a retro operating system and programming environment for ESP32 microcontrollers called BASIC32. I took it to 39% complete, as a language, over 8-weeks. Then, I spent just 50 minutes describing all the missing parts and it finished the language in 1-shot. I didn\u2019t think it would work. It did. I used Opus 5.5 on a $20 plan and I didn\u2019t even hit my limits.", "context": "", "use_case": "building a language/OS project", "switched_from": "", "story": "We should be more ambitious with the software we build"}, {"id": 803646849, "item_id": 50227928, "url": "https://news.ycombinator.com/item?id=50227928", "by": "genxy", "date": "2026-10-25", "stance": "recommend", "quote": "I used Fable 5.1 and Opus 5.5 to build a Python interpreter in Rust. Appears to pass all the 3.12 code I throw at it.", "excerpt": "I used Fable 5.1 and Opus 5.5 to build a Python interpreter in Rust. Appears to pass all the 3.12 code I throw at it. To get a toy version that can run most things takes under 2 hrs. Half of that time is collecting the tests you want to pass, how it should evaluate the code, what your criteria are, etc.", "context": "", "use_case": "building a Python interpreter", "switched_from": "", "story": "We should be more ambitious with the software we build"}, {"id": 803634384, "item_id": 50227149, "url": "https://news.ycombinator.com/item?id=50227149", "by": "Karrot_Kream", "date": "2026-10-25", "stance": "neutral", "quote": "the models can autopilot into bad decisions that are verifiably correct", "excerpt": "I used Opus 5.5 to figure out that a language model we were serving at work for internal use (we don't really serve models, we only use this internally because it processes sensitive data that we can't send to external endpoints) was being bottlenecked by CPU contention on tokenization, by having Opus run in an autoresearch style loop. Opus's first proposal for a fix was to spawn multiple Python processes and build a", "context": "works at a company serving an internal language model", "use_case": "diagnosing CPU bottleneck in model serving", "switched_from": "", "story": "We should be more ambitious with the software we build"}, {"id": 802679057, "item_id": 50167441, "url": "https://news.ycombinator.com/item?id=50167441", "by": "A_D_E_P_T", "date": "2026-10-20", "stance": "recommend", "quote": "use Opus 5.5 for review/orchestration", "excerpt": "The Google models are great \"general purpose\" and research LLMs, but they're relatively poor at coding. The way I see it, you have two options. 1) Get on the Claude Max 20x plan and use Opus 5.5 for review/orchestration and Haiku 5.5 for grunt work. Haiku 5.5 is decent, cheap, and fast. 2) Deepseek 4.1 Flash in OpenCode. You should also test MiMo v2.6. It's a little bit uneven, but might work well for whatever you're", "context": "", "use_case": "review and orchestration", "switched_from": "", "story": "Ask HN: Best value/money AI coding plan bundle?"}, {"id": 802515808, "item_id": 50157238, "url": "https://news.ycombinator.com/item?id=50157238", "by": "hollowturtle", "date": "2026-10-19", "stance": "warn", "quote": "Claude Opus sucks at architecture", "excerpt": "> We've vibecoded scripts recently that genuinely are better than what I would have written and I don't think I'm terrible. I'm ok with vibecoding scripts, I love programming the machine and making the end user happy and not dealing with bash! I don't judge scripts quality as long as they are tools in the workflow, I dont care about that, unless they will somehow be part of the production. The name Script implies tha", "context": "", "use_case": "vibecoding scripts", "switched_from": "", "story": "It's still ok to enjoy programming"}, {"id": 802107808, "item_id": 50131738, "url": "https://news.ycombinator.com/item?id=50131738", "by": "aizk", "date": "2026-10-17", "stance": "recommend", "quote": "I use Opus 5.5 and swarms of about 50 deepseek models to reverse engineer old videos games.", "excerpt": "I use Opus 5.5 and swarms of about 50 deepseek models to reverse engineer old videos games. The scale is roughly 600k loc generated from pure asm, simply by knowing what the old compiler is. Truly staggering stuff.", "context": "", "use_case": "reverse engineering old games", "switched_from": "", "story": "DeepSeek-v4.1-Flash-1017"}, {"id": 802049024, "item_id": 50128064, "url": "https://news.ycombinator.com/item?id=50128064", "by": "nl", "date": "2026-10-17", "stance": "recommend", "quote": "whereas Opus 5.5 has fixed that", "excerpt": "I agree that \"frontier model\" really does mean \"pushing the frontier of intelligence\", so Fable or Astra. They might mean the intelligence/dollar frontier aka the Pareto frontier, but honestly much of the time that's not a great measure. If you are doing repeated work on the same task (eg call center triage) then you can always beat a large LLM on price by fine tuning a small one. If you are doing knowledge work then", "context": "", "use_case": "knowledge work", "switched_from": "", "story": "Ultramarine 1\u00a0Flash"}, {"id": 801250880, "item_id": 50078180, "url": "https://news.ycombinator.com/item?id=50078180", "by": "mattlondon", "date": "2026-10-13", "stance": "warn", "quote": "I do not like the way Opus 5.5 communicates with me.", "excerpt": "Agree I do not like the way Opus 5.5 communicates with me. It feels like it is trying to act superior in a condescending way. DO NOT LIKE. I have not used Argon yet but Gemini 3.8 Flash has been a joy to use in comparison (and so fast!) as a daily driver.", "context": "", "use_case": "", "switched_from": "", "story": "Gemini 4 Argon"}, {"id": 801042912, "item_id": 50065182, "url": "https://news.ycombinator.com/item?id=50065182", "by": "9cb14c1ec0", "date": "2026-10-13", "stance": "recommend", "quote": "Opus 5.5 has built Lineage ROMs for me for 3 different devices in the last couple of weeks", "excerpt": "Opus 5.5 has built Lineage ROMs for me for 3 different devices in the last couple of weeks, that do not have official Lineage support or any one else working on them. https://github.com/9cb14c1ec0/android_device_lenovo_sycamore_row_5G-lineage https://github.com/9cb14c1ec0/android_device_tcl_selene_vzw https://github.com/9cb14c1ec0/android_device_lenovo_clove_row_wifi-lineage Quite impressive what these models are cap", "context": "", "use_case": "building Lineage ROMs", "switched_from": "", "story": "I turned my unused Kindle into a photo frame with GPT-6 Astra"}, {"id": 801026064, "item_id": 50064129, "url": "https://news.ycombinator.com/item?id=50064129", "by": "prmph", "date": "2026-10-12", "stance": "warn", "quote": "Claude Opus 5.5 still makes rudimentary mistakes all the time as I use it.", "excerpt": "I don't know. Claude Opus 5.5 still makes rudimentary mistakes all the time as I use it. More than that, it cannot go beyond a certain level. Like a dog that can only do dog-level actions. I've lost count of the number of times I step back, take a critical look at what it has been hacking away at for hours, which would seem OK, but can be replaced by a much more effective architecture. For the foreseeable future, a h", "context": "", "use_case": "coding", "switched_from": "", "story": "AI is a tool and using it well takes skill"}, {"id": 801003825, "item_id": 50062739, "url": "https://news.ycombinator.com/item?id=50062739", "by": "mattlondon", "date": "2026-10-12", "stance": "warn", "quote": "I really dislike the way it talks to me (it feels quite condescending, holier-than-thou, uptight etc)", "excerpt": "Gemini 4 Argon release a few days ago was frontier a few days ago. Things move fast. I think some of the benchmarks that Meta didn't pick show Gemini 4 Argon ahead but really the benchmarks are not that great any more and getting increasingly disconnected from reality as people search for \"headroom\" in the evals to avoid saturation I think. Opus 5.5 is \"top\" for example but I really dislike the way it talks to me (it", "context": "", "use_case": "", "switched_from": "", "story": "Muse Wonder"}]}