{"slug": "gpt", "name": "GPT", "category": "ai", "category_name": "AI models and assistants", "description": "OpenAI's family of large language models, available through ChatGPT and the OpenAI API.", "recommenders": 22, "commenters": 62, "warners": 7, "score": 21.905, "rising": 0.76, "first": "2025-11-24", "latest": "2026-10-31", "why": "HN commenters like GPT most as a code reviewer and plan checker alongside other models, and for following instructions closely without refusing.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 23, "neutral": 28, "compare": 9, "warn": 8}, "spark": [0, 0, 0, 0, 0, 1, 1, 9, 6, 3, 2, 1], "listed": true, "loved": ["Code review", "Core logic"], "caveat": "Verbose output", "short_quote": "This (among many others) is the reason I use GPT over Claude. It adheres very closely to your instructions and rules.", "icon": true, "domain": "swelljoe.com", "github_repo": null, "official_url": "https://openai.com/", "pricing": "freemium", "aliases": ["GPT"], "verdict": [{"text": "HN commenters like GPT most as a code reviewer and plan checker alongside other models,", "cites": [{"id": 791116049, "item_id": 49444753, "url": "https://news.ycombinator.com/item?id=49444753", "by": "adventured", "date": "2026-08-26", "stance": "recommend", "quote": "I use GPT for planning/roadmaps and code review, it checks Fable.", "excerpt": "I use Fable for writing code and some planning. I use GPT for planning/roadmaps and code review, it checks Fable. I use Gemini for stray conversations. Gemini has a better conversational style than GPT, less robotic; it hallucinates more unfortunately, it has plainly fallen quite a bit behind, but it's still highly useful to converse with, research with, etc. And it spares me wasting usage of GPT or Fable. I use Gemi", "context": "", "use_case": "planning and code review", "switched_from": "", "story": "Analyzing student votes across AI models for college essay help"}, {"id": 777851602, "item_id": 48615725, "url": "https://news.ycombinator.com/item?id=48615725", "by": "jimbobimbo", "date": "2026-06-20", "stance": "recommend", "quote": "then gpt review the plan and provide feedback", "excerpt": "This is the way. I use gh copilot and have opus interrogate me and write the plan, then gpt review the plan and provide feedback; repeat this multiple times until gpt is either satisfied or starts to nitpick on unimportant stuff. Then sanity check the plan myself and have gpt implement it. Each implementation is also reviewed by me before merging to master. I complete PRs only when I'm satisfied with the implementati", "context": "", "use_case": "reviewing plans, implementing", "switched_from": "", "story": "When I reject AI code even if it works"}, {"id": 768832209, "item_id": 48052013, "url": "https://news.ycombinator.com/item?id=48052013", "by": "anthonycoslett", "date": "2026-05-07", "stance": "recommend", "quote": "using GPT purely as an adversarial collaborator to spot check Claude's work", "excerpt": "I hear you. I'd be interested to learn if you have found ways to mitigate the inaccurate information? I use Claude and GPT - using GPT purely as an adversarial collaborator to spot check Claude's work (and sometimes vice versa) it sounds counterproductive and hopefully at some point that will be unnecessary but it works wonders for the final products I put the stamp of approval on. I think it's indicative of the fact", "context": "", "use_case": "spot-checking Claude's work", "switched_from": "", "story": "I Still Can't Trust AI"}, {"id": 777848721, "item_id": 48615545, "url": "https://news.ycombinator.com/item?id=48615545", "by": "wwind123", "date": "2026-06-20", "stance": "recommend", "quote": "I use 3 AI's (Claude, GPT and Gemini) to review each other's design plans and implementation on the same code base.", "excerpt": "I use 3 AI's (Claude, GPT and Gemini) to review each other's design plans and implementation on the same code base. Each often catches problems the others miss. I try to make sure the architecture docs of the code base are refreshed regularly based on recent changes, so it's easier for humans and AI agents to make sense of the code. I also regularly stop all other developments and just focus on auditing the code base", "context": "", "use_case": "cross-reviewing design and code", "switched_from": "", "story": "When I reject AI code even if it works"}]}, {"text": " and for following instructions closely without refusing.", "cites": [{"id": 784868352, "item_id": 49054272, "url": "https://news.ycombinator.com/item?id=49054272", "by": "nullbio", "date": "2026-07-25", "stance": "recommend", "quote": "This (among many others) is the reason I use GPT over Claude. It adheres very closely to your instructions and rules.", "excerpt": "We want more control, not less. This (among many others) is the reason I use GPT over Claude. It adheres very closely to your instructions and rules. You can build up your own workflows and systems as a result. Claude just does whatever it wants, regardless of what you tell it. It's a miserable experience. Claude is what you use when you want to one-shot a simple cookie-cutter product that has no nuance in it. A prod", "context": "", "use_case": "complex codebases", "switched_from": "", "story": "The new rules of context engineering for Claude 5 generation models"}, {"id": 782025842, "item_id": 48876615, "url": "https://news.ycombinator.com/item?id=48876615", "by": "Leynos", "date": "2026-07-11", "stance": "recommend", "quote": "Yes, I tend to prefer GPT", "excerpt": "Yeah, I went through a period after 4.7 launched of not using Claude for code work at all because of the condescending refusals. (Kept using it for planning and design work). Still had three pretty bad refusals from 4.8, but not the same quantity as from 4.7. Claude.ai is pretty frustrating too. It will talk to me as if anything I ask for is below it. Could be the system prompt I use, but I did not have this issue wi", "context": "", "use_case": "", "switched_from": "", "story": "I used to love Claude, but the latest models are slowly ruining it"}]}, {"text": " The most common complaint is verbose or overengineered output,", "cites": [{"id": 779158465, "item_id": 48697404, "url": "https://news.ycombinator.com/item?id=48697404", "by": "mschuetz", "date": "2026-06-27", "stance": "warn", "quote": "GPT's code ends up way more verbose/complex/overengineered than it needs to be", "excerpt": "I haven't tried the latest Codex but I switched from GPT to Claude because I think Claude writes much better Code. GPT's code ends up way more verbose/complex/overengineered than it needs to be.", "context": "", "use_case": "writing code", "switched_from": "", "story": "Previewing GPT\u20115.6 Sol: a next-generation model"}, {"id": 766716066, "item_id": 47919754, "url": "https://news.ycombinator.com/item?id=47919754", "by": "jdw64", "date": "2026-04-27", "stance": "recommend", "quote": "GPT works well as a validator. It tends to consider far more edge cases than I would.", "excerpt": "These kinds of comparisons only really make sense if the setup is controlled \u2014 same prompt, same agent configuration (like AGENT.md / CLAUDE.md), and similar usage patterns. In practice, the results vary a lot depending on how you actually use each model. In my case, I found that Claude is better for frontend design and initial structure, while GPT tends to perform better on core logic. Claude often misses basic secu", "context": "", "use_case": "strict code review", "switched_from": "", "story": "Claude Code Opus-4-7 VS Codex GPT-5-5"}, {"id": 793519683, "item_id": 49594980, "url": "https://news.ycombinator.com/item?id=49594980", "by": "ramon156", "date": "2026-09-07", "stance": "compare", "quote": "Claude/GPT really wants to append more text for some reason.", "excerpt": "Disclaimer: I hand curate them in the end I keep most of my sessions in Zed (you can import them there anyway). After some big feature I let a frontier agent go over these sessions and suggest improvements. Typically I use gemini for this because it's really good at pruning text. Claude/GPT really wants to append more text for some reason. I end up with smaller skills but more \"actioned\" skills. They kind of force th", "context": "", "use_case": "", "switched_from": "", "story": "Ask HN: How do you manage skills files?"}]}, {"text": " and some say it rarely explains its reasoning or assumptions.", "cites": [{"id": 786022496, "item_id": 49126406, "url": "https://news.ycombinator.com/item?id=49126406", "by": "jauntywundrkind", "date": "2026-07-31", "stance": "warn", "quote": "I value gpt so much, but it is such a worse peer to me than the other models I use.", "excerpt": "This is the thing! This is so so so important. It's such an outrage. I value gpt so much, but it is such a worse peer to me than the other models I use. It delivers without explaining. I can sit and ask questions that it will answer but it is not a peer, does not share readily ever. It will not tell me what assumptions it's baking in. It won't tell what directions or invariants it's trying to hold or break, what it c", "context": "", "use_case": "", "switched_from": "", "story": "The Session You Cannot take with you"}, {"id": 778659121, "item_id": 48666195, "url": "https://news.ycombinator.com/item?id=48666195", "by": "jauntywundrkind", "date": "2026-06-24", "stance": "warn", "quote": "It was so jarring to switch to GPT and Opus and find that they won't discuss with me, won't reveal their thinking", "excerpt": "I think the self-doubt might actually be a very crucial part of it's capability. I often feel compelled to interrupt when I'm watching it think (which thank the stars it let's us do, unlike the big American models!!), but usually it makes the right pick! Being willing and able to reconsider seems very good. Going around and around, pulling in more thinking, integrating it: maybe that's why it is as good as it's good.", "context": "", "use_case": "", "switched_from": "", "story": "GLM-5.2 is the step change for open agents"}]}], "praise": [{"label": "Code review", "summary": "Often used to review plans and check other models' code, catching gaps and edge cases.", "quotes": [{"id": 783854912, "item_id": 48990932, "url": "https://news.ycombinator.com/item?id=48990932", "by": "dimonb19a", "date": "2026-07-21", "stance": "recommend", "quote": "I use GPT only for audits and serious code logic because it can catch gaps and inconsistencies that Claude misses.", "excerpt": "Each model has a different personality. There is some real layer underneath but it also depends on how you use it. For me GPT is a nerd and Claude i consider like a senior dev with deep knowledge. I use GPT only for audits and serious code logic because it can catch gaps and inconsistencies that Claude misses. But it always depends on your mindset, i think what really matters is a loop - you plus AI.", "context": "", "use_case": "code audits, code logic", "switched_from": "", "story": "Ask HN: Claude Code or Codex?"}, {"id": 766716066, "item_id": 47919754, "url": "https://news.ycombinator.com/item?id=47919754", "by": "jdw64", "date": "2026-04-27", "stance": "recommend", "quote": "GPT works well as a validator. It tends to consider far more edge cases than I would.", "excerpt": "These kinds of comparisons only really make sense if the setup is controlled \u2014 same prompt, same agent configuration (like AGENT.md / CLAUDE.md), and similar usage patterns. In practice, the results vary a lot depending on how you actually use each model. In my case, I found that Claude is better for frontend design and initial structure, while GPT tends to perform better on core logic. Claude often misses basic secu", "context": "", "use_case": "strict code review", "switched_from": "", "story": "Claude Code Opus-4-7 VS Codex GPT-5-5"}, {"id": 777851602, "item_id": 48615725, "url": "https://news.ycombinator.com/item?id=48615725", "by": "jimbobimbo", "date": "2026-06-20", "stance": "recommend", "quote": "then gpt review the plan and provide feedback", "excerpt": "This is the way. I use gh copilot and have opus interrogate me and write the plan, then gpt review the plan and provide feedback; repeat this multiple times until gpt is either satisfied or starts to nitpick on unimportant stuff. Then sanity check the plan myself and have gpt implement it. Each implementation is also reviewed by me before merging to master. I complete PRs only when I'm satisfied with the implementati", "context": "", "use_case": "reviewing plans, implementing", "switched_from": "", "story": "When I reject AI code even if it works"}]}, {"label": "Core logic", "summary": "Seen as more thorough on core logic and architecture than some alternatives.", "quotes": [{"id": 776057521, "item_id": 48503595, "url": "https://news.ycombinator.com/item?id=48503595", "by": "jdw64", "date": "2026-06-12", "stance": "recommend", "quote": "GPT series models are more thorough and better", "excerpt": "Personally, when I use open code or routers, I feel that beyond a certain level, the models don't make a huge difference to me. Except for expensive and mediocre models like Gemini. In that sense, Chinese models are pretty good. I usually write code in function or method units and then design and assemble them together. GPT series models are more thorough and better, but I'm not sure if the difference is enormous. It", "context": "", "use_case": "coding", "switched_from": "", "story": "Kimi K2.7-Code: open-source coding model with better token efficiency"}, {"id": 785287585, "item_id": 49080474, "url": "https://news.ycombinator.com/item?id=49080474", "by": "solarkraft", "date": "2026-07-28", "stance": "compare", "quote": "The GPTs are better at general architecting", "excerpt": "> even the lower-end models can find the things they didn't think of first when inspecting the already-written output I use Deepseek V4 Pro for my hobby project (an OpenCode client). The economics will obviously be different at work. It\u2019s worse than the GPTs, but indeed, if you focus it on reviewing its own code (essentially spending more reasoning and changing perspective), it\u2019s also quite capable at improving its o", "context": "", "use_case": "architecting", "switched_from": "", "story": "A $500 RL fine-tune of a 9B open model beat frontier models on catalog review"}, {"id": 789580353, "item_id": 49348772, "url": "https://news.ycombinator.com/item?id=49348772", "by": "hirvi74", "date": "2026-08-18", "stance": "compare", "quote": "It appears both Claude and GPT believe GPT's solutions are better (and so I do).", "excerpt": "While ancedata does not mean much, I have had horrible success with Claude lately. I have been using Claude to crosscheck some of the outputs from GPT and vice versa. It appears both Claude and GPT believe GPT's solutions are better (and so I do). I still believe Claude has a better UI/UX in the web interface, but tolerating Anthropic's bullshit is not worth it.", "context": "", "use_case": "cross-checking outputs", "switched_from": "", "story": "Claude: Degraded Performance for Multiple Models"}]}, {"label": "Instruction following", "summary": "Praised for sticking to user rules and doing work other models refuse.", "quotes": [{"id": 784868352, "item_id": 49054272, "url": "https://news.ycombinator.com/item?id=49054272", "by": "nullbio", "date": "2026-07-25", "stance": "recommend", "quote": "This (among many others) is the reason I use GPT over Claude. It adheres very closely to your instructions and rules.", "excerpt": "We want more control, not less. This (among many others) is the reason I use GPT over Claude. It adheres very closely to your instructions and rules. You can build up your own workflows and systems as a result. Claude just does whatever it wants, regardless of what you tell it. It's a miserable experience. Claude is what you use when you want to one-shot a simple cookie-cutter product that has no nuance in it. A prod", "context": "", "use_case": "complex codebases", "switched_from": "", "story": "The new rules of context engineering for Claude 5 generation models"}, {"id": 782025842, "item_id": 48876615, "url": "https://news.ycombinator.com/item?id=48876615", "by": "Leynos", "date": "2026-07-11", "stance": "recommend", "quote": "Yes, I tend to prefer GPT", "excerpt": "Yeah, I went through a period after 4.7 launched of not using Claude for code work at all because of the condescending refusals. (Kept using it for planning and design work). Still had three pretty bad refusals from 4.8, but not the same quantity as from 4.7. Claude.ai is pretty frustrating too. It will talk to me as if anything I ask for is below it. Could be the system prompt I use, but I did not have this issue wi", "context": "", "use_case": "", "switched_from": "", "story": "I used to love Claude, but the latest models are slowly ruining it"}, {"id": 781409282, "item_id": 48838080, "url": "https://news.ycombinator.com/item?id=48838080", "by": "SwellJoe", "date": "2026-07-08", "stance": "recommend", "quote": "I recently switched to using GPT for Nelson and the security benchmarks", "excerpt": "I've found in my current work on a security auditing harness and benchmarks, both Fable and Opus are useless. I recently switched to using GPT for Nelson and the security benchmarks I've been doing because Opus started refusing to do the work. I guess I probably could also use GLM or DeepSeek or MiMo, and I'll probably do some experiments to see the shape of all of their guardrails in this area soon, now that I see i", "context": "building a security auditing harness", "use_case": "security auditing", "switched_from": "Opus", "story": "Fable is not a useful model"}]}, {"label": "Tone", "summary": "Some prefer its plainer, less confrontational tone over Claude's.", "quotes": [{"id": 776540001, "item_id": 48533750, "url": "https://news.ycombinator.com/item?id=48533750", "by": "kmac_", "date": "2026-06-14", "stance": "recommend", "quote": "OpenAI models are much more tame and to the point (I use the same prompt)", "excerpt": "It isn't new behavior. I use each model to redact emails. Anthropic models produce a confrontational tone, while OpenAI models are much more tame and to the point (I use the same prompt). I noticed that a long time ago and prefer GPT for those tasks.", "context": "", "use_case": "redacting emails", "switched_from": "Claude", "story": "Why Is Claude Turning into an a**Hole?"}, {"id": 795118768, "item_id": 49694923, "url": "https://news.ycombinator.com/item?id=49694923", "by": "r_lee", "date": "2026-09-14", "stance": "recommend", "quote": "I just switched to the GPT models and it's a breath of fresh air", "excerpt": "yeah, I just switched to the GPT models and it's a breath of fresh air. as for the other guy, the claude talk is definitely not less ambiguous, it often is incredibly ambiguous and hard to parse, I have no clue why it produces such output, if not to fingerprint it? it's really weird man. when Opus 5 came out, I was really confused. I saw a bunch of hype about how it's better than fable, but I just felt frustrated wit", "context": "", "use_case": "coding assistance", "switched_from": "Claude", "story": "Fable 5.1 Solves the Cyphral Distich, a 370-year-old cipher"}, {"id": 798161394, "item_id": 49885087, "url": "https://news.ycombinator.com/item?id=49885087", "by": "saint-evan", "date": "2026-09-28", "stance": "recommend", "quote": "I prefer talking to GPT", "excerpt": "unwittingly said 'yaaay' when I saw the Sonnet 5.5 entry. lol I love Sonnet so much. Loved it since 3.5 and never really liked Opus even when I tried to use it for technical work. Once we had two back to back anthropic releases neither of which was a Sonnet upgrade from 4.5 (4.6?) and I was getting kinda sad that they're considering discontinuing it. These are weird reactions I'm having to these tools even when I mos", "context": "", "use_case": "conversation", "switched_from": "", "story": "Sonnet 5.5"}]}, {"label": "Plan limits", "summary": "The $20/mo plan is described as enough for daily work.", "quotes": [{"id": 778350594, "item_id": 48646912, "url": "https://news.ycombinator.com/item?id=48646912", "by": "stavros", "date": "2026-06-23", "stance": "recommend", "quote": "GPT is inexhaustible on the $20/mo plan", "excerpt": "Not the GP, but I use Opus for planning, Deepseek for actual coding (implementing the plan) and GPT for review. GPT is inexhaustible on the $20/mo plan, Deepseek is dirt cheap (maybe $10/mo) and Claude is Claude.", "context": "", "use_case": "code review", "switched_from": "", "story": "AI's Affordability Crisis"}, {"id": 779419808, "item_id": 48713738, "url": "https://news.ycombinator.com/item?id=48713738", "by": "jklmnopqrstuvw", "date": "2026-06-28", "stance": "neutral", "quote": "I use GPT for months. $20/m plus plan is enough for my daily work.", "excerpt": "> A typical session for me with GPT is usually over a hundred dollars. I don't think a $100 session is \"typical\". I use GPT for months. $20/m plus plan is enough for my daily work.", "context": "", "use_case": "daily work", "switched_from": "", "story": "Semgrep: GLM 5.2 beats Claude in our Cyber Benchmarks"}]}, {"label": "Explaining concepts", "summary": "Used to simplify complex material and find learning resources.", "quotes": [{"id": 782563489, "item_id": 48910218, "url": "https://news.ycombinator.com/item?id=48910218", "by": "nextzck", "date": "2026-07-14", "stance": "recommend", "quote": "Take it to sonnet 5 or gpt and ask it to explain this to a layman.", "excerpt": "Take it to sonnet 5 or gpt and ask it to explain this to a layman. If you still don\u2019t get it ask it for the why it matters or the how it relates. You can also ask fable/4.8 to do it but I find it helps to keep the working model surrounded by the complexity rather than drawing it out. Simplifying text is something that takes relatively low effort in comparison to technical tasks. Sometimes I use Gemini, deepseek, grok", "context": "", "use_case": "simplifying technical explanations", "switched_from": "", "story": "How to stop Claude from saying load-bearing"}, {"id": 789738528, "item_id": 49358658, "url": "https://news.ycombinator.com/item?id=49358658", "by": "eudy_ai_web", "date": "2026-08-19", "stance": "recommend", "quote": "Just ask GPT or Claude to find some good courses and learning paths on GitHub.", "excerpt": "Just ask GPT or Claude to find some good courses and learning paths on GitHub. Start by building demos\u2014learn by doing.", "context": "", "use_case": "finding learning resources", "switched_from": "", "story": "Ask HN: Where to learn software engineering fundamentals?"}]}], "caveats": [{"label": "Verbose output", "summary": "Code and text can come out overengineered, overly defensive or padded.", "quotes": [{"id": 779158465, "item_id": 48697404, "url": "https://news.ycombinator.com/item?id=48697404", "by": "mschuetz", "date": "2026-06-27", "stance": "warn", "quote": "GPT's code ends up way more verbose/complex/overengineered than it needs to be", "excerpt": "I haven't tried the latest Codex but I switched from GPT to Claude because I think Claude writes much better Code. GPT's code ends up way more verbose/complex/overengineered than it needs to be.", "context": "", "use_case": "writing code", "switched_from": "", "story": "Previewing GPT\u20115.6 Sol: a next-generation model"}, {"id": 766716066, "item_id": 47919754, "url": "https://news.ycombinator.com/item?id=47919754", "by": "jdw64", "date": "2026-04-27", "stance": "recommend", "quote": "GPT works well as a validator. It tends to consider far more edge cases than I would.", "excerpt": "These kinds of comparisons only really make sense if the setup is controlled \u2014 same prompt, same agent configuration (like AGENT.md / CLAUDE.md), and similar usage patterns. In practice, the results vary a lot depending on how you actually use each model. In my case, I found that Claude is better for frontend design and initial structure, while GPT tends to perform better on core logic. Claude often misses basic secu", "context": "", "use_case": "strict code review", "switched_from": "", "story": "Claude Code Opus-4-7 VS Codex GPT-5-5"}, {"id": 793519683, "item_id": 49594980, "url": "https://news.ycombinator.com/item?id=49594980", "by": "ramon156", "date": "2026-09-07", "stance": "compare", "quote": "Claude/GPT really wants to append more text for some reason.", "excerpt": "Disclaimer: I hand curate them in the end I keep most of my sessions in Zed (you can import them there anyway). After some big feature I let a frontier agent go over these sessions and suggest improvements. Typically I use gemini for this because it's really good at pruning text. Claude/GPT really wants to append more text for some reason. I end up with smaller skills but more \"actioned\" skills. They kind of force th", "context": "", "use_case": "", "switched_from": "", "story": "Ask HN: How do you manage skills files?"}]}, {"label": "Hidden reasoning", "summary": "Some say it delivers without sharing assumptions or thinking.", "quotes": [{"id": 786022496, "item_id": 49126406, "url": "https://news.ycombinator.com/item?id=49126406", "by": "jauntywundrkind", "date": "2026-07-31", "stance": "warn", "quote": "I value gpt so much, but it is such a worse peer to me than the other models I use.", "excerpt": "This is the thing! This is so so so important. It's such an outrage. I value gpt so much, but it is such a worse peer to me than the other models I use. It delivers without explaining. I can sit and ask questions that it will answer but it is not a peer, does not share readily ever. It will not tell me what assumptions it's baking in. It won't tell what directions or invariants it's trying to hold or break, what it c", "context": "", "use_case": "", "switched_from": "", "story": "The Session You Cannot take with you"}, {"id": 778659121, "item_id": 48666195, "url": "https://news.ycombinator.com/item?id=48666195", "by": "jauntywundrkind", "date": "2026-06-24", "stance": "warn", "quote": "It was so jarring to switch to GPT and Opus and find that they won't discuss with me, won't reveal their thinking", "excerpt": "I think the self-doubt might actually be a very crucial part of it's capability. I often feel compelled to interrupt when I'm watching it think (which thank the stars it let's us do, unlike the big American models!!), but usually it makes the right pick! Being willing and able to reconsider seems very good. Going around and around, pulling in more thinking, integrating it: maybe that's why it is as good as it's good.", "context": "", "use_case": "", "switched_from": "", "story": "GLM-5.2 is the step change for open agents"}]}, {"label": "Accuracy errors", "summary": "Users report repeated arithmetic mistakes and poor results in Elixir.", "quotes": [{"id": 799482176, "item_id": 49967636, "url": "https://news.ycombinator.com/item?id=49967636", "by": "Forgeties79", "date": "2026-10-05", "stance": "warn", "quote": "GPT messed up the math for me the other day when I was simply adding 10 durations for a TRT.", "excerpt": "> I use \"AI\" a lot, it is a fantastic tool. We need to stop pretending it is some kind of panacea. It's a tool. I have been repeating some variation of this for months. If people would stop acting like it\u2019s THE tech solution to ALL things ALL the time I bet a lot of critics would quiet down. The overhyping has become exhausting. It\u2019s been going on for years. GPT messed up the math for me the other day when I was simp", "context": "", "use_case": "arithmetic", "switched_from": "", "story": "The Philadelphia Inquirer built Scrape, an AI tool to surface hyperlocal news"}, {"id": 783418512, "item_id": 48963657, "url": "https://news.ycombinator.com/item?id=48963657", "by": "rubyn00bie", "date": "2026-07-18", "stance": "warn", "quote": "I have found GPT to be pretty awful at writing Elixir", "excerpt": "I think it depends on the problem space, and to a large extent the languages used. I have had the opposite experience with GPT. I have found GPT to be pretty awful at writing Elixir, and dealing with more open ended tasks. Opus is (for the moment) better at writing Elixir, and Fable is much better than either at actually grokking a problem space. I had switched to GPT for about six weeks and eventually stopped using ", "context": "Elixir developer", "use_case": "Elixir coding", "switched_from": "", "story": "Fable 5 vs. GPT-5.6 Sol on an NP-Hard Problem: Does /goal help?"}]}, {"label": "Prose writing", "summary": "One commenter finds it unacceptable for writing English text.", "quotes": [{"id": 784700034, "item_id": 49043752, "url": "https://news.ycombinator.com/item?id=49043752", "by": "hodgehog11", "date": "2026-07-24", "stance": "warn", "quote": "Neither Claude nor GPT are acceptable for writing English text.", "excerpt": "Neither Claude nor GPT are acceptable for writing English text. Personally I have found Gemini to be far better, and that is really all I use it for.", "context": "", "use_case": "writing English text", "switched_from": "", "story": "Claude Opus 5"}]}, {"label": "Refusals", "summary": "One user reports it refused a task it considered illegal.", "quotes": [{"id": 776653217, "item_id": 48540826, "url": "https://news.ycombinator.com/item?id=48540826", "by": "LaurensBER", "date": "2026-06-15", "stance": "warn", "quote": "GPT refused to do so (citing that it's illegal even though I own the games).", "excerpt": "Same, I had Deepseek search for, download and transfer (to my Linux emulation machine) the best Dreamcast games yesterday. GPT refused to do so (citing that it's illegal even though I own the games). Deepseek did a wonderful job for 7 cents. At work I use Opus because, why not? But I could easily switch to a less capable model if needed.", "context": "", "use_case": "", "switched_from": "", "story": "Anthropic's Safety Superpower"}]}], "who": [{"by": "Jash271", "context": "not a video editor", "url": "https://news.ycombinator.com/item?id=50311807", "date": "2026-10-31"}, {"by": "boltzmann-brain", "context": "Senior software engineer", "url": "https://news.ycombinator.com/item?id=49929994", "date": "2026-10-02"}, {"by": "woctordho", "context": "user in China", "url": "https://news.ycombinator.com/item?id=49544132", "date": "2026-09-02"}, {"by": "meliz242", "context": "wrote a personal Substack investigation", "url": "https://news.ycombinator.com/item?id=49515215", "date": "2026-08-31"}, {"by": "insanitybit", "context": "M4 MacBook user", "url": "https://news.ycombinator.com/item?id=49149033", "date": "2026-08-02"}, {"by": "AbbeFaria", "context": "doing a distributed systems course", "url": "https://news.ycombinator.com/item?id=48917573", "date": "2026-07-15"}, {"by": "iugtmkbdfil834", "context": "not an active trader", "url": "https://news.ycombinator.com/item?id=48881820", "date": "2026-07-12"}, {"by": "SwellJoe", "context": "building a security auditing harness", "url": "https://news.ycombinator.com/item?id=48838080", "date": "2026-07-08"}, {"by": "Philip-J-Fry", "context": "developer who uses Chinese models for personal projects", "url": "https://news.ycombinator.com/item?id=48828487", "date": "2026-07-07"}, {"by": "jdw64", "context": "developer who delivers finished products", "url": "https://news.ycombinator.com/item?id=48044151", "date": "2026-05-06"}, {"by": "eddiehammond", "context": "hiring for Kepler", "url": "https://news.ycombinator.com/item?id=47999884", "date": "2026-05-03"}], "tags": ["for code review", "for multi-model workflows", "for planning and design docs", "for security auditing"], "use_cases": ["coding", "character images", "porting key bindings", "answering daily published questions", "arithmetic", "day-to-day development"], "compared": [{"slug": "claude", "name": "Claude", "co": 51, "switched_from": 3, "switched_to": 1}, {"slug": "claude-opus", "name": "Claude Opus", "co": 21, "switched_from": 1, "switched_to": 1}, {"slug": "gemini", "name": "Gemini", "co": 20, "switched_from": 0, "switched_to": 1}, {"slug": "glm", "name": "GLM", "co": 9, "switched_from": 0, "switched_to": 0}, {"slug": "deepseek", "name": "DeepSeek", "co": 9, "switched_from": 0, "switched_to": 0}, {"slug": "claude-fable", "name": "Claude Fable", "co": 8, "switched_from": 0, "switched_to": 0}, {"slug": "cursor", "name": "Cursor", "co": 7, "switched_from": 0, "switched_to": 0}, {"slug": "codex", "name": "Codex", "co": 6, "switched_from": 0, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=49527595", "title": "Path to Astra: critical capabilities and frontier safeguards", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=49443871", "title": "Analyzing student votes across AI models for college essay help", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=49290713", "title": "Ask HN: How much money do you spend monthly on subscriptions for AI models?", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=48956879", "title": "Fable 5 vs. GPT-5.6 Sol on an NP-Hard Problem: Does /goal help?", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=48614631", "title": "When I reject AI code even if it works", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=50311801", "title": "Where AI agents' social boundaries break: 5 LLMs try to form a government", "mentions": 1}], "sentiment_by_month": {"2025-11": {"neutral": 1}, "2025-12": {}, "2026-01": {}, "2026-02": {"compare": 1}, "2026-03": {"compare": 2}, "2026-04": {"recommend": 1}, "2026-05": {"neutral": 4, "recommend": 1}, "2026-06": {"recommend": 9, "neutral": 2, "warn": 3}, "2026-07": {"neutral": 6, "recommend": 6, "warn": 3, "compare": 1}, "2026-08": {"neutral": 8, "warn": 1, "compare": 2, "recommend": 3}, "2026-09": {"neutral": 4, "compare": 3, "recommend": 2}, "2026-10": {"neutral": 3, "warn": 1, "recommend": 1}}, "quotes": [{"id": 804988914, "item_id": 50311807, "url": "https://news.ycombinator.com/item?id=50311807", "by": "Jash271", "date": "2026-10-31", "stance": "neutral", "quote": "GPT for the images of the characters", "excerpt": "Agents interacting with each other made for great content, so we created a series out of it: https://www.youtube.com/watch?v=dxXvgM75yis PS: I'm not a video editor. I used Claude to generate the videos, ElevenLabs for the voice, and GPT for the images of the characters", "context": "not a video editor", "use_case": "character images", "switched_from": "", "story": "Where AI agents' social boundaries break: 5 LLMs try to form a government"}, {"id": 804762402, "item_id": 50297650, "url": "https://news.ycombinator.com/item?id=50297650", "by": "Jash271", "date": "2026-10-29", "stance": "neutral", "quote": "GPT for the images of the characters", "excerpt": "Agents interacting with each other made for great content, so we created a series out of it: https://www.youtube.com/watch?v=dxXvgM75yis PS: I'm not a video editor. I used Claude to generate the videos, ElevenLabs for the voice, and GPT for the images of the characters.", "context": "", "use_case": "character images", "switched_from": "", "story": "Where AI agents' social boundaries break: 5 LLMs try to form a government"}, {"id": 804379444, "item_id": 50273715, "url": "https://news.ycombinator.com/item?id=50273715", "by": "vr46", "date": "2026-10-28", "stance": "recommend", "quote": "I did ask GPT to port over my key commands and shortcuts, which it did, and things are great again!", "excerpt": "About two or three months ago I ditched Neovim as all the complicated modern crap wasn't working for me, and instead of reverting to Vim, I switched to Emacs and Doom + eVil, and am relieved to be back in a nice stable system again. I did ask GPT to port over my key commands and shortcuts, which it did, and things are great again!", "context": "", "use_case": "porting key bindings", "switched_from": "", "story": "Emacs Turns 50"}, {"id": 801014720, "item_id": 50063420, "url": "https://news.ycombinator.com/item?id=50063420", "by": "QuestionClass", "date": "2026-10-12", "stance": "neutral", "quote": "then I\u2019d throw it into GPT and post the answer", "excerpt": "Most of my favorite meetings/interactions/conversations have come with great questions. I love walking away thinking, \u201cwhat a great question, I see that from a new perspective.\u201d Or at least an additional. So I started publishing one question a day, then I\u2019d throw it into GPT and post the answer. To date I\u2019ve published 842 questions, after I write the question and think about it for a minute or two, maybe jot down a t", "context": "", "use_case": "answering daily published questions", "switched_from": "", "story": null}, {"id": 799482176, "item_id": 49967636, "url": "https://news.ycombinator.com/item?id=49967636", "by": "Forgeties79", "date": "2026-10-05", "stance": "warn", "quote": "GPT messed up the math for me the other day when I was simply adding 10 durations for a TRT.", "excerpt": "> I use \"AI\" a lot, it is a fantastic tool. We need to stop pretending it is some kind of panacea. It's a tool. I have been repeating some variation of this for months. If people would stop acting like it\u2019s THE tech solution to ALL things ALL the time I bet a lot of critics would quiet down. The overhyping has become exhausting. It\u2019s been going on for years. GPT messed up the math for me the other day when I was simp", "context": "", "use_case": "arithmetic", "switched_from": "", "story": "The Philadelphia Inquirer built Scrape, an AI tool to surface hyperlocal news"}, {"id": 798879904, "item_id": 49929994, "url": "https://news.ycombinator.com/item?id=49929994", "by": "boltzmann-brain", "date": "2026-10-02", "stance": "neutral", "quote": "I use GPT, Codex, Claude, Cursor and other current tools extensively.", "excerpt": "Location: Austria, EU Remote: Yes, US/EU/UK/Global Technologies: Rust, Python, Go, Haskell, C++, TypeScript, JS, React, K8s, PHP, ... Code: https://github.com/cheater R\u00e9sum\u00e9/CV: ask Email: damianjobsites+hn@gmail.com AI-native Senior Software Engineer with 10+ years of hands-on experience building difficult systems in AI, fintech, high-performance APIs and crypto. \u2022 Part of original Tezos launch team - blockchain pro", "context": "Senior software engineer", "use_case": "day-to-day development", "switched_from": "", "story": "Ask HN: Who wants to be hired? (October 2026)"}, {"id": 798161394, "item_id": 49885087, "url": "https://news.ycombinator.com/item?id=49885087", "by": "saint-evan", "date": "2026-09-28", "stance": "recommend", "quote": "I prefer talking to GPT", "excerpt": "unwittingly said 'yaaay' when I saw the Sonnet 5.5 entry. lol I love Sonnet so much. Loved it since 3.5 and never really liked Opus even when I tried to use it for technical work. Once we had two back to back anthropic releases neither of which was a Sonnet upgrade from 4.5 (4.6?) and I was getting kinda sad that they're considering discontinuing it. These are weird reactions I'm having to these tools even when I mos", "context": "", "use_case": "conversation", "switched_from": "", "story": "Sonnet 5.5"}, {"id": 796715728, "item_id": 49794733, "url": "https://news.ycombinator.com/item?id=49794733", "by": "daquisu", "date": "2026-09-21", "stance": "neutral", "quote": "I pay for GPT, Claude and Gemini.", "excerpt": "One possible reason is that Grok Bot consumes more quota because it uses their infra instead of your computer. I can't help much more than that, I did that research in the last few days, but I never used Grok myself. I pay for GPT, Claude and Gemini. Last week I consumed all my quota on two of them, so I wondered which next subscription I would pay for if needed.", "context": "", "use_case": "AI assistant subscription", "switched_from": "", "story": "Grok 4.7"}, {"id": 795118768, "item_id": 49694923, "url": "https://news.ycombinator.com/item?id=49694923", "by": "r_lee", "date": "2026-09-14", "stance": "recommend", "quote": "I just switched to the GPT models and it's a breath of fresh air", "excerpt": "yeah, I just switched to the GPT models and it's a breath of fresh air. as for the other guy, the claude talk is definitely not less ambiguous, it often is incredibly ambiguous and hard to parse, I have no clue why it produces such output, if not to fingerprint it? it's really weird man. when Opus 5 came out, I was really confused. I saw a bunch of hype about how it's better than fable, but I just felt frustrated wit", "context": "", "use_case": "coding assistance", "switched_from": "Claude", "story": "Fable 5.1 Solves the Cyphral Distich, a 370-year-old cipher"}, {"id": 794420416, "item_id": 49651276, "url": "https://news.ycombinator.com/item?id=49651276", "by": "mohamedkoubaa", "date": "2026-09-10", "stance": "compare", "quote": "I find GPT to be better at understanding physics conceptually but Claude is better at writing numerical code.", "excerpt": "I'm in the same field and I find GPT to be better at understanding physics conceptually but Claude is better at writing numerical code. I use cursor so many of my sessions start in GPT and switch to Claude.", "context": "works in simulation software", "use_case": "physics conceptual reasoning", "switched_from": "", "story": "Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra"}, {"id": 793519683, "item_id": 49594980, "url": "https://news.ycombinator.com/item?id=49594980", "by": "ramon156", "date": "2026-09-07", "stance": "compare", "quote": "Claude/GPT really wants to append more text for some reason.", "excerpt": "Disclaimer: I hand curate them in the end I keep most of my sessions in Zed (you can import them there anyway). After some big feature I let a frontier agent go over these sessions and suggest improvements. Typically I use gemini for this because it's really good at pruning text. Claude/GPT really wants to append more text for some reason. I end up with smaller skills but more \"actioned\" skills. They kind of force th", "context": "", "use_case": "", "switched_from": "", "story": "Ask HN: How do you manage skills files?"}, {"id": 792706112, "item_id": 49544132, "url": "https://news.ycombinator.com/item?id=49544132", "by": "woctordho", "date": "2026-09-02", "stance": "neutral", "quote": "that's how exactly I use GPT and Claude in China", "excerpt": "There are 'transfer stations' and that's how exactly I use GPT and Claude in China. OpenAI and Anthropic do not sell in China, so we use their AI with a much lower price like 1% of the official API price. The largest transfer stations have TBs of traffic every day, and the traffic is eventually possessed by the open source community. Subscription engineering is a deep field. Neither OpenAI nor Anthropic have any tech", "context": "user in China", "use_case": "via transfer stations", "switched_from": "", "story": "Path to Astra: critical capabilities and frontier safeguards"}, {"id": 792579682, "item_id": 49536230, "url": "https://news.ycombinator.com/item?id=49536230", "by": "not_a_bot_4sho", "date": "2026-09-02", "stance": "neutral", "quote": "Sticking to a mix of GPT and Gemini for now.", "excerpt": "I use GHCP but similar sentiment. Stopped using Anthropic models for this reason. Their prose become too obtuse and just... alien. No human talks or writes like that. It's incredibly taxing to deal with. Sticking to a mix of GPT and Gemini for now.", "context": "", "use_case": "", "switched_from": "Claude", "story": "Claude Fable 5.1 and Claude Mythos 5.1"}, {"id": 792546512, "item_id": 49534157, "url": "https://news.ycombinator.com/item?id=49534157", "by": "disgruntledphd2", "date": "2026-09-02", "stance": "compare", "quote": "I find it hard to distinguish between the outputs of GPT/Claude/Kimi/GLM recently", "excerpt": "Where does the secret sauce show up in the outputs, then? Like, (apart from tone), I find it hard to distinguish between the outputs of GPT/Claude/Kimi/GLM recently (I use cursor, and have been giving them the same prompt and comparing). If anything, I found that the non-Claude models were better in many cases, which definitely doesn't map to their pricing. > in fact not all the methods and data are public Probably n", "context": "", "use_case": "coding prompts", "switched_from": "", "story": "Path to Astra: critical capabilities and frontier safeguards"}, {"id": 792302673, "item_id": 49518917, "url": "https://news.ycombinator.com/item?id=49518917", "by": "terhechte", "date": "2026-09-01", "stance": "neutral", "quote": "I'm using GPT (Terra, Sol, Luna), Claude (Opus 5, Fable), Qwen 3.8 and GLM 5.3 Flash daily", "excerpt": "It is the point. Also open source model enthusiast tell you otherwise, there is a coding quality gap between these models. If I use DeepSeek, I do so knowing that I have to limit to simpler tasks on smaller, well specified prompts. What the author did, letting the model do the planning, is not something DeepSeek will excel at. I'm using GPT (Terra, Sol, Luna), Claude (Opus 5, Fable), Qwen 3.8 and GLM 5.3 Flash daily ", "context": "", "use_case": "coding", "switched_from": "", "story": "AI Can Make You Suck Faster Too"}, {"id": 792243440, "item_id": 49515215, "url": "https://news.ycombinator.com/item?id=49515215", "by": "meliz242", "date": "2026-08-31", "stance": "neutral", "quote": "I eventually asked gpt if I was correct in my understanding about refrigeration controllers", "excerpt": "Nope, I'm one of those lucky folks who was privileged enough to be accused of writing in a weird and probably too mature for my high school English paper-way before the advent of gpt, lol. I've had to all but remove the em-dash from my vernacular, thanks to the ~ plague of inauthenticity ~ as I call it. I did all the digging, researching, writing, etc myself. The reason the screenshots are from mobile is because I wr", "context": "wrote a personal Substack investigation", "use_case": "research questions", "switched_from": "", "story": "I think the military commissary's freezers were hacked"}, {"id": 792218688, "item_id": 49513668, "url": "https://news.ycombinator.com/item?id=49513668", "by": "tptacek", "date": "2026-08-31", "stance": "recommend", "quote": "feeding them through GPT and Claude one by one, and it's drastically increased the speed at which I can knock out a completed piece", "excerpt": "I appreciate what this piece is trying to do directionally but it shares a logical flaw with lots of other pieces about AI and knowledge work. It assumes that in order to disrupt a particular field, AI needs to print serviceable work unassisted; that it's \"vibe-shipping\". AI today is most effective when it's not vibing, but rather copiloting a skilled operator. I don't necessarily want my agent to build an entire sys", "context": "", "use_case": "editing feedback on writing", "switched_from": "", "story": "The safest job from AI may be writing"}, {"id": 791188595, "item_id": 49449287, "url": "https://news.ycombinator.com/item?id=49449287", "by": "spobin", "date": "2026-08-26", "stance": "neutral", "quote": "Four of the eight clubs are managed by frontier AI models (Claude, GPT, Gemini, Manus).", "excerpt": "I've been fascinated by Unitree's G1 robot architecture and, without the $20k to buy a unit, decided to play around with the MuJoCo rendering system. I fell down a bit of a rabbit hole and ended up with an automated football league that runs 3 matches per day and streams on Twitch/Youtube. It's been a lot of fun! Four of the eight clubs are managed by frontier AI models (Claude, GPT, Gemini, Manus). Between match day", "context": "", "use_case": "managing football clubs", "switched_from": "", "story": "Show HN: A robot football league where frontier AI models manage the clubs"}, {"id": 791116049, "item_id": 49444753, "url": "https://news.ycombinator.com/item?id=49444753", "by": "adventured", "date": "2026-08-26", "stance": "recommend", "quote": "I use GPT for planning/roadmaps and code review, it checks Fable.", "excerpt": "I use Fable for writing code and some planning. I use GPT for planning/roadmaps and code review, it checks Fable. I use Gemini for stray conversations. Gemini has a better conversational style than GPT, less robotic; it hallucinates more unfortunately, it has plainly fallen quite a bit behind, but it's still highly useful to converse with, research with, etc. And it spares me wasting usage of GPT or Fable. I use Gemi", "context": "", "use_case": "planning and code review", "switched_from": "", "story": "Analyzing student votes across AI models for college essay help"}, {"id": 791113936, "item_id": 49444621, "url": "https://news.ycombinator.com/item?id=49444621", "by": "gexla", "date": "2026-08-26", "stance": "compare", "quote": "GPT gets the most out there, but creates interesting nuance that turn an idea into a more creative exercise.", "excerpt": "GPT gets the most out there, but creates interesting nuance that turn an idea into a more creative exercise. Claude is like a more grounded GPT but may miss the nuance. Gemini is my last pick, but does better than the rest for making something clear and understandable. It gets to be exhausting parsing through Claude outputs and breaking it down into something more easily understandable. One way I use to improve this ", "context": "", "use_case": "idea exploration", "switched_from": "", "story": "Analyzing student votes across AI models for college essay help"}, {"id": 790349474, "item_id": 49396842, "url": "https://news.ycombinator.com/item?id=49396842", "by": "irchans", "date": "2026-08-22", "stance": "neutral", "quote": "I asked GPT and it suggested that I use ficl", "excerpt": "For years I've had an idea in my head to add a Forth to the old game Master of Orion so that I could automate some of my common actions in the game. There is an open source GIThub called 1oom with source code for the game and I thought I would just write my own Forth. Well, I've recently been using AI's to code, so I asked GPT and it suggested that I use ficl rather than writing my own Forth. It helped me add ficl to", "context": "", "use_case": "coding help", "switched_from": "", "story": "Tumble Forth \u2013 from assembly to OS with C compiler (2023)"}, {"id": 789844769, "item_id": 49365298, "url": "https://news.ycombinator.com/item?id=49365298", "by": "ramoz", "date": "2026-08-19", "stance": "neutral", "quote": "I was actually using it more with GPT models but they have been getting better.", "excerpt": "I created /bro a couple model iterations prior. I use it everyday many times a day and it's getting worse. I was actually using it more with GPT models but they have been getting better. https://github.com/backnotprop/bro/blob/main/skills/bro/SKILL.md", "context": "", "use_case": "running the /bro skill", "switched_from": "", "story": "Opus 5.0 drives incoherence into the stratosphere"}, {"id": 789738528, "item_id": 49358658, "url": "https://news.ycombinator.com/item?id=49358658", "by": "eudy_ai_web", "date": "2026-08-19", "stance": "recommend", "quote": "Just ask GPT or Claude to find some good courses and learning paths on GitHub.", "excerpt": "Just ask GPT or Claude to find some good courses and learning paths on GitHub. Start by building demos\u2014learn by doing.", "context": "", "use_case": "finding learning resources", "switched_from": "", "story": "Ask HN: Where to learn software engineering fundamentals?"}, {"id": 789580353, "item_id": 49348772, "url": "https://news.ycombinator.com/item?id=49348772", "by": "hirvi74", "date": "2026-08-18", "stance": "compare", "quote": "It appears both Claude and GPT believe GPT's solutions are better (and so I do).", "excerpt": "While ancedata does not mean much, I have had horrible success with Claude lately. I have been using Claude to crosscheck some of the outputs from GPT and vice versa. It appears both Claude and GPT believe GPT's solutions are better (and so I do). I still believe Claude has a better UI/UX in the web interface, but tolerating Anthropic's bullshit is not worth it.", "context": "", "use_case": "cross-checking outputs", "switched_from": "", "story": "Claude: Degraded Performance for Multiple Models"}, {"id": 789520049, "item_id": 49345003, "url": "https://news.ycombinator.com/item?id=49345003", "by": "SkyBelow", "date": "2026-08-18", "stance": "warn", "quote": "I've seen the same with GPT models, with Grok, with Deepseek", "excerpt": "This isn't just on code. When I'm writing technical documentation, it keeps the explanations in. Same when writing non-technical documentation. When I was having it attempt to generate a Pathfinder 1e class for a Sword Dancer, it was leaving in notes about why it removes things I told it to remove/rework. And it isn't just Claude. I've seen the same with GPT models, with Grok, with Deepseek. Each AI isn't quite the s", "context": "", "use_case": "writing documentation and code", "switched_from": "", "story": "AI;DR (AI; Didn't Read)"}, {"id": 788931938, "item_id": 49308246, "url": "https://news.ycombinator.com/item?id=49308246", "by": "dexterlagan", "date": "2026-08-14", "stance": "neutral", "quote": "my usual GPT/Claude/DeepSeek for code", "excerpt": "Tested the model briefly with my usual eval: a couple of questions on general knowledge most small models often get wrong, then write a fully-featured todo list app in JS, then rewrite the same app in Rust with Tauri. Granted, most models are well trained on basic todo apps, but it gives me an idea of the basic SWE capabilities I can build on. As far as I'm concerned, if it can successfully setup a local git repo, wr", "context": "", "use_case": "coding", "switched_from": "", "story": "Qwen 3.8 27B is out: open weights, best local dense model yet"}, {"id": 788732931, "item_id": 49295808, "url": "https://news.ycombinator.com/item?id=49295808", "by": "davedx", "date": "2026-08-14", "stance": "neutral", "quote": "I do also use the API's for GPT and Gemini for some side projects", "excerpt": "I have subscriptions for ChatGPT Plus and Claude Pro, so about $40/month. I use them for coding mostly, switching between them frequently and heavily. On the occasional times I run out of tokens I'll go outside and take a walk or spend time with family. Been doing it this way for ~the last 6 months or so. I do also use the API's for GPT and Gemini for some side projects, but the spend there is highly variable. It rar", "context": "", "use_case": "side projects via API", "switched_from": "", "story": "Ask HN: How much money do you spend monthly on subscriptions for AI models?"}, {"id": 788653057, "item_id": 49290816, "url": "https://news.ycombinator.com/item?id=49290816", "by": "Join_TalkoryAi", "date": "2026-08-13", "stance": "neutral", "quote": "I naturally use several models\u2014GPT, Claude, Gemini, Grok, Perplexity, etc.", "excerpt": "For me, the biggest pain point was actually having multiple AI subscriptions just to use different models for different things. I\u2019m working on Talkory.ai, so I naturally use several models\u2014GPT, Claude, Gemini, Grok, Perplexity, etc.\u2014for things like coding, research, content and brainstorming. Individually, each subscription makes sense, but collectively the monthly cost adds up pretty quickly. That\u2019s one of the reaso", "context": "", "use_case": "coding, research, content", "switched_from": "", "story": "Ask HN: How much money do you spend monthly on subscriptions for AI models?"}, {"id": 786384529, "item_id": 49149033, "url": "https://news.ycombinator.com/item?id=49149033", "by": "insanitybit", "date": "2026-08-02", "stance": "neutral", "quote": "I use Codex, gpt, and Claude in various configurations", "excerpt": "I'm on an M4 Macbook. I use Codex, gpt, and Claude in various configurations, including Fable, xhigh, medium, high, ultracode, etc.", "context": "M4 MacBook user", "use_case": "coding with LLMs", "switched_from": "", "story": "(Rust) All-Hands 2026 Retrospective"}, {"id": 786022496, "item_id": 49126406, "url": "https://news.ycombinator.com/item?id=49126406", "by": "jauntywundrkind", "date": "2026-07-31", "stance": "warn", "quote": "I value gpt so much, but it is such a worse peer to me than the other models I use.", "excerpt": "This is the thing! This is so so so important. It's such an outrage. I value gpt so much, but it is such a worse peer to me than the other models I use. It delivers without explaining. I can sit and ask questions that it will answer but it is not a peer, does not share readily ever. It will not tell me what assumptions it's baking in. It won't tell what directions or invariants it's trying to hold or break, what it c", "context": "", "use_case": "", "switched_from": "", "story": "The Session You Cannot take with you"}, {"id": 785287585, "item_id": 49080474, "url": "https://news.ycombinator.com/item?id=49080474", "by": "solarkraft", "date": "2026-07-28", "stance": "compare", "quote": "The GPTs are better at general architecting", "excerpt": "> even the lower-end models can find the things they didn't think of first when inspecting the already-written output I use Deepseek V4 Pro for my hobby project (an OpenCode client). The economics will obviously be different at work. It\u2019s worse than the GPTs, but indeed, if you focus it on reviewing its own code (essentially spending more reasoning and changing perspective), it\u2019s also quite capable at improving its o", "context": "", "use_case": "architecting", "switched_from": "", "story": "A $500 RL fine-tune of a 9B open model beat frontier models on catalog review"}, {"id": 784868352, "item_id": 49054272, "url": "https://news.ycombinator.com/item?id=49054272", "by": "nullbio", "date": "2026-07-25", "stance": "recommend", "quote": "This (among many others) is the reason I use GPT over Claude. It adheres very closely to your instructions and rules.", "excerpt": "We want more control, not less. This (among many others) is the reason I use GPT over Claude. It adheres very closely to your instructions and rules. You can build up your own workflows and systems as a result. Claude just does whatever it wants, regardless of what you tell it. It's a miserable experience. Claude is what you use when you want to one-shot a simple cookie-cutter product that has no nuance in it. A prod", "context": "", "use_case": "complex codebases", "switched_from": "", "story": "The new rules of context engineering for Claude 5 generation models"}, {"id": 784700034, "item_id": 49043752, "url": "https://news.ycombinator.com/item?id=49043752", "by": "hodgehog11", "date": "2026-07-24", "stance": "warn", "quote": "Neither Claude nor GPT are acceptable for writing English text.", "excerpt": "Neither Claude nor GPT are acceptable for writing English text. Personally I have found Gemini to be far better, and that is really all I use it for.", "context": "", "use_case": "writing English text", "switched_from": "", "story": "Claude Opus 5"}, {"id": 783960290, "item_id": 48997518, "url": "https://news.ycombinator.com/item?id=48997518", "by": "adventured", "date": "2026-07-21", "stance": "neutral", "quote": "you can just pay as you go with GPT and Fable", "excerpt": "The Chinese have been right behind OpenAI and Anthropic for ~18 months now. DeepSeek didn't do to OpenAI and Anthropic what nearly everybody claimed they would. Every single person on HN that loudly proclaimed the end was nigh for GPT & Co. due to DeepSeek, was wrong. They were humiliatingly wrong, and they'll never own up to it. The reason those people were so very wrong, is the same exact reason the Kimi crowd is w", "context": "", "use_case": "pay-as-you-go model access", "switched_from": "", "story": "Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber"}, {"id": 783854912, "item_id": 48990932, "url": "https://news.ycombinator.com/item?id=48990932", "by": "dimonb19a", "date": "2026-07-21", "stance": "recommend", "quote": "I use GPT only for audits and serious code logic because it can catch gaps and inconsistencies that Claude misses.", "excerpt": "Each model has a different personality. There is some real layer underneath but it also depends on how you use it. For me GPT is a nerd and Claude i consider like a senior dev with deep knowledge. I use GPT only for audits and serious code logic because it can catch gaps and inconsistencies that Claude misses. But it always depends on your mindset, i think what really matters is a loop - you plus AI.", "context": "", "use_case": "code audits, code logic", "switched_from": "", "story": "Ask HN: Claude Code or Codex?"}, {"id": 783418512, "item_id": 48963657, "url": "https://news.ycombinator.com/item?id=48963657", "by": "rubyn00bie", "date": "2026-07-18", "stance": "warn", "quote": "I have found GPT to be pretty awful at writing Elixir", "excerpt": "I think it depends on the problem space, and to a large extent the languages used. I have had the opposite experience with GPT. I have found GPT to be pretty awful at writing Elixir, and dealing with more open ended tasks. Opus is (for the moment) better at writing Elixir, and Fable is much better than either at actually grokking a problem space. I had switched to GPT for about six weeks and eventually stopped using ", "context": "Elixir developer", "use_case": "Elixir coding", "switched_from": "", "story": "Fable 5 vs. GPT-5.6 Sol on an NP-Hard Problem: Does /goal help?"}, {"id": 783319969, "item_id": 48957498, "url": "https://news.ycombinator.com/item?id=48957498", "by": "o10449366", "date": "2026-07-18", "stance": "recommend", "quote": "Even just spending 10 minutes forcing GPT to write a design doc results in much more robust plans than plan mode", "excerpt": "/goal has replaced plan mode for me. This is the pattern I use for 95% of my AI work now: 1. Read X feature of Y and tell me when you fully understand it (if there's any detail missing in the summary, repeat until the context is primed) 2. What time is it? 3. /goal Spend X minutes from $time writing a technical design doc on $feature. There must not be any vague language or ambiguity in the document. Read carry_forwa", "context": "", "use_case": "writing design docs", "switched_from": "", "story": "Fable 5 vs. GPT-5.6 Sol on an NP-Hard Problem: Does /goal help?"}, {"id": 782879185, "item_id": 48929949, "url": "https://news.ycombinator.com/item?id=48929949", "by": "edg5000", "date": "2026-07-15", "stance": "neutral", "quote": "I do about 10 google search queries for every 1 opus/gpt prompt", "excerpt": "I do about 10 google search queries for every 1 opus/gpt prompt. For google, I don't actually open pages anymore 9 out 10 times; I rely on the AI summary. It's fast and accurate; the trick is that you learn where the boundary is of what you can ask it. Querying information the small model is great at. Then there might be slow, batch tasks. I can see myself getting 1T of slow RAM one day (in a few years?) and having a", "context": "", "use_case": "", "switched_from": "", "story": "Running Gemma 4 26B at 5 tokens/SEC on a 13-year-old Xeon with no GPU"}, {"id": 782681169, "item_id": 48917573, "url": "https://news.ycombinator.com/item?id=48917573", "by": "AbbeFaria", "date": "2026-07-15", "stance": "neutral", "quote": "using Claude/GPT to ask for references etc is good enough for me", "excerpt": "I haven\u2019t watched the lectures yet. Reading the papers and using Claude/GPT to ask for references etc is good enough for me. I did watch a few, ones on GFS and Raft but didn\u2019t see the ROI, lectures are too long. Reading the papers multiple times and reading blogs, references and asking Q\u2019s to GPT is what I am doing. Obviously for the assignments themselves, I use AI sparingly. Mostly as a sounding board when I have d", "context": "doing a distributed systems course", "use_case": "studying distributed systems papers", "switched_from": "", "story": "Paxos Made Simple (2001)[pdf]"}, {"id": 782563489, "item_id": 48910218, "url": "https://news.ycombinator.com/item?id=48910218", "by": "nextzck", "date": "2026-07-14", "stance": "recommend", "quote": "Take it to sonnet 5 or gpt and ask it to explain this to a layman.", "excerpt": "Take it to sonnet 5 or gpt and ask it to explain this to a layman. If you still don\u2019t get it ask it for the why it matters or the how it relates. You can also ask fable/4.8 to do it but I find it helps to keep the working model surrounded by the complexity rather than drawing it out. Simplifying text is something that takes relatively low effort in comparison to technical tasks. Sometimes I use Gemini, deepseek, grok", "context": "", "use_case": "simplifying technical explanations", "switched_from": "", "story": "How to stop Claude from saying load-bearing"}, {"id": 782025842, "item_id": 48876615, "url": "https://news.ycombinator.com/item?id=48876615", "by": "Leynos", "date": "2026-07-11", "stance": "recommend", "quote": "Yes, I tend to prefer GPT", "excerpt": "Yeah, I went through a period after 4.7 launched of not using Claude for code work at all because of the condescending refusals. (Kept using it for planning and design work). Still had three pretty bad refusals from 4.8, but not the same quantity as from 4.7. Claude.ai is pretty frustrating too. It will talk to me as if anything I ask for is below it. Could be the system prompt I use, but I did not have this issue wi", "context": "", "use_case": "", "switched_from": "", "story": "I used to love Claude, but the latest models are slowly ruining it"}, {"id": 779158465, "item_id": 48697404, "url": "https://news.ycombinator.com/item?id=48697404", "by": "mschuetz", "date": "2026-06-27", "stance": "warn", "quote": "GPT's code ends up way more verbose/complex/overengineered than it needs to be", "excerpt": "I haven't tried the latest Codex but I switched from GPT to Claude because I think Claude writes much better Code. GPT's code ends up way more verbose/complex/overengineered than it needs to be.", "context": "", "use_case": "writing code", "switched_from": "", "story": "Previewing GPT\u20115.6 Sol: a next-generation model"}, {"id": 778659121, "item_id": 48666195, "url": "https://news.ycombinator.com/item?id=48666195", "by": "jauntywundrkind", "date": "2026-06-24", "stance": "warn", "quote": "It was so jarring to switch to GPT and Opus and find that they won't discuss with me, won't reveal their thinking", "excerpt": "I think the self-doubt might actually be a very crucial part of it's capability. I often feel compelled to interrupt when I'm watching it think (which thank the stars it let's us do, unlike the big American models!!), but usually it makes the right pick! Being willing and able to reconsider seems very good. Going around and around, pulling in more thinking, integrating it: maybe that's why it is as good as it's good.", "context": "", "use_case": "", "switched_from": "", "story": "GLM-5.2 is the step change for open agents"}, {"id": 777851602, "item_id": 48615725, "url": "https://news.ycombinator.com/item?id=48615725", "by": "jimbobimbo", "date": "2026-06-20", "stance": "recommend", "quote": "then gpt review the plan and provide feedback", "excerpt": "This is the way. I use gh copilot and have opus interrogate me and write the plan, then gpt review the plan and provide feedback; repeat this multiple times until gpt is either satisfied or starts to nitpick on unimportant stuff. Then sanity check the plan myself and have gpt implement it. Each implementation is also reviewed by me before merging to master. I complete PRs only when I'm satisfied with the implementati", "context": "", "use_case": "reviewing plans, implementing", "switched_from": "", "story": "When I reject AI code even if it works"}, {"id": 777848721, "item_id": 48615545, "url": "https://news.ycombinator.com/item?id=48615545", "by": "wwind123", "date": "2026-06-20", "stance": "recommend", "quote": "I use 3 AI's (Claude, GPT and Gemini) to review each other's design plans and implementation on the same code base.", "excerpt": "I use 3 AI's (Claude, GPT and Gemini) to review each other's design plans and implementation on the same code base. Each often catches problems the others miss. I try to make sure the architecture docs of the code base are refreshed regularly based on recent changes, so it's easier for humans and AI agents to make sense of the code. I also regularly stop all other developments and just focus on auditing the code base", "context": "", "use_case": "cross-reviewing design and code", "switched_from": "", "story": "When I reject AI code even if it works"}, {"id": 768832209, "item_id": 48052013, "url": "https://news.ycombinator.com/item?id=48052013", "by": "anthonycoslett", "date": "2026-05-07", "stance": "recommend", "quote": "using GPT purely as an adversarial collaborator to spot check Claude's work", "excerpt": "I hear you. I'd be interested to learn if you have found ways to mitigate the inaccurate information? I use Claude and GPT - using GPT purely as an adversarial collaborator to spot check Claude's work (and sometimes vice versa) it sounds counterproductive and hopefully at some point that will be unnecessary but it works wonders for the final products I put the stamp of approval on. I think it's indicative of the fact", "context": "", "use_case": "spot-checking Claude's work", "switched_from": "", "story": "I Still Can't Trust AI"}, {"id": 766716066, "item_id": 47919754, "url": "https://news.ycombinator.com/item?id=47919754", "by": "jdw64", "date": "2026-04-27", "stance": "recommend", "quote": "GPT works well as a validator. It tends to consider far more edge cases than I would.", "excerpt": "These kinds of comparisons only really make sense if the setup is controlled \u2014 same prompt, same agent configuration (like AGENT.md / CLAUDE.md), and similar usage patterns. In practice, the results vary a lot depending on how you actually use each model. In my case, I found that Claude is better for frontend design and initial structure, while GPT tends to perform better on core logic. Claude often misses basic secu", "context": "", "use_case": "strict code review", "switched_from": "", "story": "Claude Code Opus-4-7 VS Codex GPT-5-5"}]}