{"slug": "minimax", "name": "MiniMax", "category": "ai", "category_name": "AI models and assistants", "description": "AI company offering MiniMax large language models via API, coding subscriptions and open weights.", "recommenders": 17, "commenters": 24, "warners": 2, "score": 13.777, "rising": -0.58, "first": "2026-02-12", "latest": "2026-10-23", "why": "HN commenters like MiniMax for very cheap token plans and pay-as-you-go use and as a capable model for agents, subagents and utility tasks like code exploration.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 19, "neutral": 6, "compare": 2, "warn": 2}, "spark": [0, 0, 0, 0, 2, 3, 0, 5, 4, 3, 1, 1], "listed": true, "loved": ["Low cost", "Agentic tasks"], "caveat": "Coding quality", "short_quote": "I've been very surprised with MiniMax M3, not as a particularly good architect, but at it's very good ability to state the problem elegantly", "icon": true, "domain": null, "github_repo": null, "official_url": "https://www.minimax.io", "pricing": "freemium", "aliases": ["MiniMax", "MiniMax M3", "MiniMax M2.7", "Minimax M3", "MiniMax M2.5", "Minimax M2.5", "MiniMax-m2.7", "MiniMax-M3"], "verdict": [{"text": "HN commenters like MiniMax for very cheap token plans and pay-as-you-go use", "cites": [{"id": 790685683, "item_id": 49417855, "url": "https://news.ycombinator.com/item?id=49417855", "by": "dkersten", "date": "2026-08-24", "stance": "recommend", "quote": "a MiniMax subscription for high-token-use tasks that don\u2019t need the best model (for $20 I get what feels like infinite tokens)", "excerpt": "I stopped using Claude because of this BS. If I pay for a service, I want to know what I\u2019m paying for, I want it to be predictable. Anthropic have been anything but. Flip flopping on model availability, model access behind an opaque filter, their past behaviour of model degradation as they prepared their next model\u2026 these are not signs of a reliable service. I\u2019ve mostly settled on using a mixture of open weights mode", "context": "", "use_case": "high-token-use tasks", "switched_from": "Claude", "story": "Anthropic's best AI model struggles to attract users as cheaper tools thrive"}, {"id": 777714768, "item_id": 48607173, "url": "https://news.ycombinator.com/item?id=48607173", "by": "throwa356262", "date": "2026-06-20", "stance": "recommend", "quote": "Currently testing M3 for agentic tasks. It works OK and their token plan is very cheap. Highly recommend for claw / hermes type of work.", "excerpt": "Since I quit my Claude subscription, every month I spend $20 (the cost of CC pro plan) playing around with new models and new providers. Currently testing M3 for agentic tasks. It works OK and their token plan is very cheap. Highly recommend for claw / hermes type of work. Tested GLM 5.1 for coding last month and it burned through my tokens a bit too quickly, but it worked well enough. Don't seem to have a coding pla", "context": "", "use_case": "agentic tasks", "switched_from": "Claude", "story": "MiniMax M3 vs. GLM 5.2: Codegen comparison across autonomous coding tasks"}, {"id": 763550688, "item_id": 47721918, "url": "https://news.ycombinator.com/item?id=47721918", "by": "cat_plus_plus", "date": "2026-04-10", "stance": "recommend", "quote": "I just got MiniMax $200/year token plan. Usually it works fine for daily coding", "excerpt": "I just got MiniMax $200/year token plan. Usually it works fine for daily coding, if it gets stuck I pay for some Claude API calls through Roo gateway. Unlike other plans, this one officially supports running OpenClaw or other API workflows and doesn't suspend you long term if you use too many tokens, just set rate per few hours.", "context": "", "use_case": "daily coding", "switched_from": "", "story": "Reallocating $100/Month Claude Code Spend to Zed and OpenRouter"}, {"id": 767437764, "item_id": 47964860, "url": "https://news.ycombinator.com/item?id=47964860", "by": "alexjplant", "date": "2026-04-30", "stance": "recommend", "quote": "pay as you go on cheaper models like MiniMax you can get a lot done for not a lot of coin", "excerpt": "> casting about for viable egress paths if I need to flee Claude Code Check out OpenCode (the OSS product [1]) and OpenCode Go/Zen (the LLMaaS [2]). Use a more expensive model with larger context (like GLM-5.1) for orchestration and cheaper models for coding and iteration on acceptance criteria (writing and passing tests). I also throw a more expensive vision-capable model into the mix like Gemini 3 Flash to iterate ", "context": "", "use_case": "cheap coding iteration", "switched_from": "", "story": "Claude Code refuses requests or charges extra if your commits mention \"OpenClaw\""}]}, {"text": " and as a capable model for agents, subagents and utility tasks like code exploration.", "cites": [{"id": 794842179, "item_id": 49677636, "url": "https://news.ycombinator.com/item?id=49677636", "by": "ricardobeat", "date": "2026-09-12", "stance": "recommend", "quote": "My current go-tos are DS Flash, Minimax M3 for subagents.", "excerpt": "Opus 5 at low effort is quite good at most tasks, and will go a long way even on the base plan. At higher thinking levels it is unbearably cautious and verbose - it drives me mad. I don\u2019t have a use for Fable at current prices. My current go-tos are DS Flash, Minimax M3 for subagents. Adding Muse Spark 1.3 to that mix but not convinced yet.", "context": "", "use_case": "subagents", "switched_from": "", "story": "Ask HN: What default model do you use and why?"}, {"id": 785312919, "item_id": 49082057, "url": "https://news.ycombinator.com/item?id=49082057", "by": "Frannky", "date": "2026-07-28", "stance": "recommend", "quote": "MiniMax M3 for agents", "excerpt": "I think they're doing something wrong. I have access to better models, but I just use 4.6 and 4.8. Fable and Opus 5 burn tokens too fast for me. I also switched to a cheaper plan to work less and to experiment with OpenCode and open models when I hit the 5h limits. I'm doing this partly because I feel they've already started the enshittification of the old models, and I want to be able to just stop paying for the sub", "context": "", "use_case": "agents", "switched_from": "", "story": "\"Opus 5 is a really bad model\""}, {"id": 781066306, "item_id": 48816644, "url": "https://news.ycombinator.com/item?id=48816644", "by": "jeremyjh", "date": "2026-07-07", "stance": "recommend", "quote": "I configured it to use MiniMax M3 for this task", "excerpt": "If you use a good harness or add the right tools and plugins both the image and web search issues mentioned are non-issues. oh-my-pi (omp.sh) handles images for text models out of the box - as long as you have any vision capable provider enabled, it will be used when you paste images to a text model. Rather than let it guess I configured it to use MiniMax M3 for this task (as well as other utility tasks like code exp", "context": "", "use_case": "image handling and utility tasks", "switched_from": "", "story": "GLM 5.2 and the coming AI margin collapse"}, {"id": 778082049, "item_id": 48630128, "url": "https://news.ycombinator.com/item?id=48630128", "by": "jeremyjh", "date": "2026-06-22", "stance": "recommend", "quote": "I also use MiniMax-M3 in utility roles like explore/library tasks.", "excerpt": "This is the way but Anthropic doesn\u2019t make it easy, so I use GPT 5.5 in that role since I can use my subscription in OpenCode or OMP. I also use MiniMax-M3 in utility roles like explore/library tasks. I\u2019ve had a z.ai subscription for several months so I\u2019m on the older pricing. I\u2019m really not sure it would make sense to do this at current rates - I could bump my Codex plan instead.", "context": "", "use_case": "explore/library tasks", "switched_from": "", "story": "GLM 5.2 vs. Opus"}]}, {"text": " The most common caveat is that it trails Claude and GLM on demanding coding work,", "cites": [{"id": 763814912, "item_id": 47738432, "url": "https://news.ycombinator.com/item?id=47738432", "by": "fg137", "date": "2026-04-12", "stance": "warn", "quote": "the quality is really far from Claude models.", "excerpt": "What's people's experience of using MiniMax for coding? I had a really bad time with it. I use (real) Claude Code for work so I know what a good model feels like. MiniMax's token plan is nice but the quality is really far from Claude models. I needed to constantly \"remind\" it to get things done. Even for a four sentence prompt in a session that is well below the context window, MiniMax would ignore half of it. This h", "context": "uses Claude Code for work", "use_case": "coding", "switched_from": "", "story": "MiniMax M2.7 Is Now Open Source"}, {"id": 783560962, "item_id": 48972560, "url": "https://news.ycombinator.com/item?id=48972560", "by": "p1necone", "date": "2026-07-19", "stance": "neutral", "quote": "Minimax M3 feels okay but definitely a step down from GLM-5.2.", "excerpt": "I use GLM-5.2 as an orchestrator model which delegates to Deepseek subagents (v4 flash or pro depending on complexity) and it works pretty well for a quite complex compiler codebase when I do deep enough up front planning for features/fixes. If you believe the benchmarks Deepseek v4 is pretty shitty at long running work in large codebases, but really really good at self contained algorithmic/math reasoning - which is", "context": "", "use_case": "fallback orchestrator", "switched_from": "", "story": "Qwen 3.8 Max Preview"}]}, {"text": " and its large models are hard to self-host on consumer hardware.", "cites": [{"id": 753243153, "item_id": 47077697, "url": "https://news.ycombinator.com/item?id=47077697", "by": "sosodev", "date": "2026-02-19", "stance": "warn", "quote": "I suspected that MiniMax M2.5 might be a bit too much for this board.", "excerpt": "Thanks for the additional info. I suspected that MiniMax M2.5 might be a bit too much for this board. 230B-A10B is just a lot to ask of the 395+ even with aggressive quantization. Particularly when you consider that the model is going to spend a lot of tokens thinking and that will eat into the comparatively smaller context window. I switched from the Unsloth 4-bit quant of Qwen3 Coder Next to the official 4-bit quan", "context": "running local models on a 395+ board", "use_case": "local agentic coding", "switched_from": "", "story": "Why AI Velocity Is Becoming a Debt Accelerator"}, {"id": 776704400, "item_id": 48544025, "url": "https://news.ycombinator.com/item?id=48544025", "by": "_davide_", "date": "2026-06-15", "stance": "neutral", "quote": "i used to mix remote and local minimax 2.7(q3) on my strix halo", "excerpt": "i used to mix remote and local minimax 2.7(q3) on my strix halo, it run at 30 tg and 220 tokens pp... it was a bit painful slow, but it was a good feeling i could stay offline. unfortunately m3 which is at opus .8 levels is 460b parameters and doesn't even fit in 128gb of memory, let alone a big context. strix halo feels like a toy for ai purposes. https://kyuz0.github.io/amd-strix-halo-toolboxes/", "context": "", "use_case": "local coding model", "switched_from": "", "story": "Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?"}, {"id": 758058291, "item_id": 47378643, "url": "https://news.ycombinator.com/item?id=47378643", "by": "Cytoplast3528", "date": "2026-03-14", "stance": "recommend", "quote": "I think only Claude Sonnet/Opus, GPT 5.2+, Minimax M2.5 are useful.", "excerpt": "I think only Claude Sonnet/Opus, GPT 5.2+, Minimax M2.5 are useful. They are all nearly impossible to self-host, unfortunately.", "context": "", "use_case": "", "switched_from": "", "story": "Ask HN: How do you use local LLMs productively?"}]}], "praise": [{"label": "Low cost", "summary": "Token plans and pay-as-you-go use are seen as cheap for heavy token use.", "quotes": [{"id": 790685683, "item_id": 49417855, "url": "https://news.ycombinator.com/item?id=49417855", "by": "dkersten", "date": "2026-08-24", "stance": "recommend", "quote": "a MiniMax subscription for high-token-use tasks that don\u2019t need the best model (for $20 I get what feels like infinite tokens)", "excerpt": "I stopped using Claude because of this BS. If I pay for a service, I want to know what I\u2019m paying for, I want it to be predictable. Anthropic have been anything but. Flip flopping on model availability, model access behind an opaque filter, their past behaviour of model degradation as they prepared their next model\u2026 these are not signs of a reliable service. I\u2019ve mostly settled on using a mixture of open weights mode", "context": "", "use_case": "high-token-use tasks", "switched_from": "Claude", "story": "Anthropic's best AI model struggles to attract users as cheaper tools thrive"}, {"id": 777714768, "item_id": 48607173, "url": "https://news.ycombinator.com/item?id=48607173", "by": "throwa356262", "date": "2026-06-20", "stance": "recommend", "quote": "Currently testing M3 for agentic tasks. It works OK and their token plan is very cheap. Highly recommend for claw / hermes type of work.", "excerpt": "Since I quit my Claude subscription, every month I spend $20 (the cost of CC pro plan) playing around with new models and new providers. Currently testing M3 for agentic tasks. It works OK and their token plan is very cheap. Highly recommend for claw / hermes type of work. Tested GLM 5.1 for coding last month and it burned through my tokens a bit too quickly, but it worked well enough. Don't seem to have a coding pla", "context": "", "use_case": "agentic tasks", "switched_from": "Claude", "story": "MiniMax M3 vs. GLM 5.2: Codegen comparison across autonomous coding tasks"}, {"id": 763550688, "item_id": 47721918, "url": "https://news.ycombinator.com/item?id=47721918", "by": "cat_plus_plus", "date": "2026-04-10", "stance": "recommend", "quote": "I just got MiniMax $200/year token plan. Usually it works fine for daily coding", "excerpt": "I just got MiniMax $200/year token plan. Usually it works fine for daily coding, if it gets stuck I pay for some Claude API calls through Roo gateway. Unlike other plans, this one officially supports running OpenClaw or other API workflows and doesn't suspend you long term if you use too many tokens, just set rate per few hours.", "context": "", "use_case": "daily coding", "switched_from": "", "story": "Reallocating $100/Month Claude Code Spend to Zed and OpenRouter"}]}, {"label": "Agentic tasks", "summary": "Used for subagents, agents and utility roles like exploration and library tasks.", "quotes": [{"id": 794842179, "item_id": 49677636, "url": "https://news.ycombinator.com/item?id=49677636", "by": "ricardobeat", "date": "2026-09-12", "stance": "recommend", "quote": "My current go-tos are DS Flash, Minimax M3 for subagents.", "excerpt": "Opus 5 at low effort is quite good at most tasks, and will go a long way even on the base plan. At higher thinking levels it is unbearably cautious and verbose - it drives me mad. I don\u2019t have a use for Fable at current prices. My current go-tos are DS Flash, Minimax M3 for subagents. Adding Muse Spark 1.3 to that mix but not convinced yet.", "context": "", "use_case": "subagents", "switched_from": "", "story": "Ask HN: What default model do you use and why?"}, {"id": 778082049, "item_id": 48630128, "url": "https://news.ycombinator.com/item?id=48630128", "by": "jeremyjh", "date": "2026-06-22", "stance": "recommend", "quote": "I also use MiniMax-M3 in utility roles like explore/library tasks.", "excerpt": "This is the way but Anthropic doesn\u2019t make it easy, so I use GPT 5.5 in that role since I can use my subscription in OpenCode or OMP. I also use MiniMax-M3 in utility roles like explore/library tasks. I\u2019ve had a z.ai subscription for several months so I\u2019m on the older pricing. I\u2019m really not sure it would make sense to do this at current rates - I could bump my Codex plan instead.", "context": "", "use_case": "explore/library tasks", "switched_from": "", "story": "GLM 5.2 vs. Opus"}, {"id": 764018611, "item_id": 47751163, "url": "https://news.ycombinator.com/item?id=47751163", "by": "Robelkidin", "date": "2026-04-13", "stance": "recommend", "quote": "MiniMAX and Claude models tend to be better AI models at doing agentic tasks.", "excerpt": "I have been experimenting with openclaw for the past 3 months and I am Software Engineer and I had difficulties setting it up and managing it. I almost bought Mac Studio to run it, bought into the hype. I do believe it absolutely amazing product. I recently launched AlitaGPT.COM (Alita.com was already registered) to provide a managed openclaw with hundreds of custom application integration. I used GCP GKE -Autopilot ", "context": "software engineer", "use_case": "agentic tasks", "switched_from": "", "story": "OpenAI GPT worst AI GPT/model vs. Claude/MiniAAX"}]}, {"label": "Output quality", "summary": "Users report surprisingly good results, with less nonsense than some rivals.", "quotes": [{"id": 787894672, "item_id": 49243417, "url": "https://news.ycombinator.com/item?id=49243417", "by": "notjes", "date": "2026-08-10", "stance": "recommend", "quote": "I have been using MiniMax M3 and I am really surprised by the quality this model spits out.", "excerpt": "Since the announcement of DeepSeek price hike, I have been using MiniMax M3 and I am really surprised by the quality this model spits out. Subjectively speaking, the bullshit MM3M produces is waaay less than DS0731. Could it be a result of less hallucinations than DS0731? https://artificialanalysis.ai/evaluations/omniscience#aa-omniscience-hallucination-rate Maybe hallucination is good for prototyping and creative wo", "context": "", "use_case": "coding", "switched_from": "DeepSeek", "story": "DeepSeek costs OpenCode Go user $1.14/day; dual DGX breaks even in 24 years"}, {"id": 779437793, "item_id": 48714862, "url": "https://news.ycombinator.com/item?id=48714862", "by": "uluckydev", "date": "2026-06-28", "stance": "recommend", "quote": "When M3 came out, I started using it, and it was actually really good.", "excerpt": "I used Claude a lot, but with Claude Code it takes a lot of context window, and it's very pricey, to be honest. Then I shifted towards Minimax. I used the coding plan because it's cheaper, but it still gets the job done. When M3 came out, I started using it, and it was actually really good. After that, I shifted towards OpenCode for my AI agent, and that's been really good as well. The best thing I realized is that i", "context": "", "use_case": "AI coding", "switched_from": "Claude Code", "story": "Semgrep: GLM 5.2 beats Claude in our Cyber Benchmarks"}, {"id": 766544707, "item_id": 47909044, "url": "https://news.ycombinator.com/item?id=47909044", "by": "DeathArrow", "date": "2026-04-26", "stance": "recommend", "quote": "I used Claude Code with GLM 5.1, MiniMax M2.7, Kimi K2.6 and had pretty good results.", "excerpt": ">I initially drafted this before my last post on how Claude Code is getting worse. I'm putting it out now so I can reference it in a future post on OpenCode. IMO, what is getting worse is not Claude Code, the CLI tool, but Anthropic API. That's what most people experience. I used Claude Code with GLM 5.1, MiniMax M2.7, Kimi K2.6 and had pretty good results. I prefer Claude Code over OpenCode because most plugins and ", "context": "", "use_case": "coding with Claude Code", "switched_from": "", "story": "It's OK to Use Agentic to Revive the Projects You Never Were Going to Finish"}]}, {"label": "Generous limits", "summary": "The plan supports API workflows and rate-limits instead of suspending heavy users.", "quotes": [{"id": 763550688, "item_id": 47721918, "url": "https://news.ycombinator.com/item?id=47721918", "by": "cat_plus_plus", "date": "2026-04-10", "stance": "recommend", "quote": "I just got MiniMax $200/year token plan. Usually it works fine for daily coding", "excerpt": "I just got MiniMax $200/year token plan. Usually it works fine for daily coding, if it gets stuck I pay for some Claude API calls through Roo gateway. Unlike other plans, this one officially supports running OpenClaw or other API workflows and doesn't suspend you long term if you use too many tokens, just set rate per few hours.", "context": "", "use_case": "daily coding", "switched_from": "", "story": "Reallocating $100/Month Claude Code Spend to Zed and OpenRouter"}, {"id": 790685683, "item_id": 49417855, "url": "https://news.ycombinator.com/item?id=49417855", "by": "dkersten", "date": "2026-08-24", "stance": "recommend", "quote": "a MiniMax subscription for high-token-use tasks that don\u2019t need the best model (for $20 I get what feels like infinite tokens)", "excerpt": "I stopped using Claude because of this BS. If I pay for a service, I want to know what I\u2019m paying for, I want it to be predictable. Anthropic have been anything but. Flip flopping on model availability, model access behind an opaque filter, their past behaviour of model degradation as they prepared their next model\u2026 these are not signs of a reliable service. I\u2019ve mostly settled on using a mixture of open weights mode", "context": "", "use_case": "high-token-use tasks", "switched_from": "Claude", "story": "Anthropic's best AI model struggles to attract users as cheaper tools thrive"}]}, {"label": "Fluent writing", "summary": "One user enjoyed talking with it for its grounded world view and fluent write-ups.", "quotes": [{"id": 802557504, "item_id": 50159844, "url": "https://news.ycombinator.com/item?id=50159844", "by": "jauntywundrkind", "date": "2026-10-20", "stance": "recommend", "quote": "I really really enjoyed talking with MiniMax M3 a lot.", "excerpt": "Skepticism is due! Good! The numbers don't mean much! Yes! I personally enjoy seeing how different models view problems. Take big plans from big models, then see how other agents reinterpret a stance. See which directions they go. I love the diversity of views they produce. It's not really relevant what the numbers are. I really really enjoyed talking with MiniMax M3 a lot. And M2.7. it had a good world view, wrote t", "context": "", "use_case": "exploring problems and plans", "switched_from": "", "story": "MiniMax M3.1 Flash"}, {"id": 802557505, "item_id": 50159844, "url": "https://news.ycombinator.com/item?id=50159844", "by": "jauntywundrkind", "date": "2026-10-20", "stance": "recommend", "quote": "And M2.7. it had a good world view", "excerpt": "Skepticism is due! Good! The numbers don't mean much! Yes! I personally enjoy seeing how different models view problems. Take big plans from big models, then see how other agents reinterpret a stance. See which directions they go. I love the diversity of views they produce. It's not really relevant what the numbers are. I really really enjoyed talking with MiniMax M3 a lot. And M2.7. it had a good world view, wrote t", "context": "", "use_case": "exploring problems and plans", "switched_from": "", "story": "MiniMax M3.1 Flash"}]}, {"label": "Transparent provider", "summary": "Recommended as a transparent provider for direct coding subscriptions.", "quotes": [{"id": 783385042, "item_id": 48961565, "url": "https://news.ycombinator.com/item?id=48961565", "by": "ignoramous", "date": "2026-07-18", "stance": "recommend", "quote": "subscribing (directly [0]) with providers that are transparent like MiniMax, DeepSeek, Xiaomi, & Z.ai.", "excerpt": "> tried Kimi K3 on a task I've done with every other model I use regularly and found it chewed a lot longer on the problem and ate up almost the entirety of a 5 hour usage limit on their $19 plan ArtificialAnalysis puts Kimi K3 just below DeepSeek v4 & GLM 5.2 in token use per task, which is about 2x to 3x more tokens than Grok 4.5: https://x.com/ArtificialAnlys/status/2077832879187620192 > Subscription usage limits ", "context": "", "use_case": "coding subscription", "switched_from": "", "story": "The Kimi K3 Moment"}, {"id": 779827250, "item_id": 48739203, "url": "https://news.ycombinator.com/item?id=48739203", "by": "Imustaskforhelp", "date": "2026-06-30", "stance": "recommend", "quote": "Z.Ai subscription/ Kimi subscription / MiniMax subscriptions are good", "excerpt": "I think that there are some subscriptions to go by. Z.ai subscription might still be interesting. I once haggled with kimi to get it for 1$ per month. I can only help in providing pointers: If you wish to go Non-API but rather subscription route: Z.Ai subscription/ Kimi subscription / MiniMax subscriptions are good. You could also take a look at ollama subscription and opencode subscriptions. If you wish to go API ro", "context": "", "use_case": "cheap coding model subscription", "switched_from": "", "story": "Claude Code Is Steganographically Marking Requests"}]}], "caveats": [{"label": "Coding quality", "summary": "Some find it far behind Claude, ignoring parts of prompts, and a step down from GLM-5.2.", "quotes": [{"id": 763814912, "item_id": 47738432, "url": "https://news.ycombinator.com/item?id=47738432", "by": "fg137", "date": "2026-04-12", "stance": "warn", "quote": "the quality is really far from Claude models.", "excerpt": "What's people's experience of using MiniMax for coding? I had a really bad time with it. I use (real) Claude Code for work so I know what a good model feels like. MiniMax's token plan is nice but the quality is really far from Claude models. I needed to constantly \"remind\" it to get things done. Even for a four sentence prompt in a session that is well below the context window, MiniMax would ignore half of it. This h", "context": "uses Claude Code for work", "use_case": "coding", "switched_from": "", "story": "MiniMax M2.7 Is Now Open Source"}, {"id": 783560962, "item_id": 48972560, "url": "https://news.ycombinator.com/item?id=48972560", "by": "p1necone", "date": "2026-07-19", "stance": "neutral", "quote": "Minimax M3 feels okay but definitely a step down from GLM-5.2.", "excerpt": "I use GLM-5.2 as an orchestrator model which delegates to Deepseek subagents (v4 flash or pro depending on complexity) and it works pretty well for a quite complex compiler codebase when I do deep enough up front planning for features/fixes. If you believe the benchmarks Deepseek v4 is pretty shitty at long running work in large codebases, but really really good at self contained algorithmic/math reasoning - which is", "context": "", "use_case": "fallback orchestrator", "switched_from": "", "story": "Qwen 3.8 Max Preview"}]}, {"label": "Self-hosting", "summary": "The models are large, so local runs are slow, need aggressive quantization, or don't fit.", "quotes": [{"id": 753243153, "item_id": 47077697, "url": "https://news.ycombinator.com/item?id=47077697", "by": "sosodev", "date": "2026-02-19", "stance": "warn", "quote": "I suspected that MiniMax M2.5 might be a bit too much for this board.", "excerpt": "Thanks for the additional info. I suspected that MiniMax M2.5 might be a bit too much for this board. 230B-A10B is just a lot to ask of the 395+ even with aggressive quantization. Particularly when you consider that the model is going to spend a lot of tokens thinking and that will eat into the comparatively smaller context window. I switched from the Unsloth 4-bit quant of Qwen3 Coder Next to the official 4-bit quan", "context": "running local models on a 395+ board", "use_case": "local agentic coding", "switched_from": "", "story": "Why AI Velocity Is Becoming a Debt Accelerator"}, {"id": 776704400, "item_id": 48544025, "url": "https://news.ycombinator.com/item?id=48544025", "by": "_davide_", "date": "2026-06-15", "stance": "neutral", "quote": "i used to mix remote and local minimax 2.7(q3) on my strix halo", "excerpt": "i used to mix remote and local minimax 2.7(q3) on my strix halo, it run at 30 tg and 220 tokens pp... it was a bit painful slow, but it was a good feeling i could stay offline. unfortunately m3 which is at opus .8 levels is 460b parameters and doesn't even fit in 128gb of memory, let alone a big context. strix halo feels like a toy for ai purposes. https://kyuz0.github.io/amd-strix-halo-toolboxes/", "context": "", "use_case": "local coding model", "switched_from": "", "story": "Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?"}, {"id": 777904162, "item_id": 48619010, "url": "https://news.ycombinator.com/item?id=48619010", "by": "pet_the_bird", "date": "2026-06-21", "stance": "recommend", "quote": "it felt like such a breath of fresh air when I used the reference full-size model", "excerpt": "Highy anecdotal: I have tried various self-hosted models using both vllm and llama.cpp. I am in a situation where I have access to large amount of memory (~320 GB). While experimenting with quantization I found that there is a non-trivial tradeoff between quality and memory footprint. Overall my experience follows the reported pattern of \"2-bit is mwah, 4-bit half decent and 6-bit required for programming. Still, alt", "context": "has access to ~320 GB memory", "use_case": "programming", "switched_from": "", "story": "Two Qwen3 models on one DGX Spark: the residency math"}]}, {"label": "Context costs", "summary": "Pay-per-token use gets pricey as the context fills up.", "quotes": [{"id": 763837187, "item_id": 47739824, "url": "https://news.ycombinator.com/item?id=47739824", "by": "rzkyif", "date": "2026-04-12", "stance": "compare", "quote": "It's quite cheap and seems to be better than Gemini 3 Flash", "excerpt": "My personal experience is way different: I struggle to burn through more than 50% of the 5 hour limit For context, with Google AI Pro, I can burn through the Antigravity weekly limit in 1-2 hours if I force it to use Gemini 3.1 Pro. Meanwhile Gemini 3 Flash is basically unlimited but frequently produces buggy code or fail to implement things how I personally would (felt like it doesn't \"think\" like a software dev) I ", "context": "", "use_case": "coding via Cline", "switched_from": "", "story": "Pro Max 5x Quota Exhausted in 1.5 Hours Despite Moderate Usage"}]}], "who": [{"by": "pet_the_bird", "context": "has access to ~320 GB memory", "url": "https://news.ycombinator.com/item?id=48619010", "date": "2026-06-21"}, {"by": "Robelkidin", "context": "software engineer", "url": "https://news.ycombinator.com/item?id=47751163", "date": "2026-04-13"}, {"by": "radio879", "context": "works with LLMs daily for 8-15 hours", "url": "https://news.ycombinator.com/item?id=47318434", "date": "2026-03-09"}], "tags": ["for agents and subagents", "for budget coding", "works with Claude Code and OpenCode", "self-hostable on big hardware"], "use_cases": ["exploring problems and plans", "coding", "agentic tasks", "implementation after planning", "exploration and library roles", "subagents"], "compared": [{"slug": "opencode", "name": "OpenCode", "co": 13, "switched_from": 0, "switched_to": 0}, {"slug": "claude-code", "name": "Claude Code", "co": 10, "switched_from": 1, "switched_to": 0}, {"slug": "deepseek", "name": "DeepSeek", "co": 9, "switched_from": 1, "switched_to": 0}, {"slug": "glm", "name": "GLM", "co": 10, "switched_from": 0, "switched_to": 0}, {"slug": "claude", "name": "Claude", "co": 5, "switched_from": 2, "switched_to": 0}, {"slug": "kimi", "name": "Kimi", "co": 6, "switched_from": 1, "switched_to": 0}, {"slug": "codex", "name": "Codex", "co": 5, "switched_from": 0, "switched_to": 0}, {"slug": "openrouter", "name": "OpenRouter", "co": 5, "switched_from": 0, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=50154540", "title": "MiniMax M3.1 Flash", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=49242728", "title": "DeepSeek costs OpenCode Go user $1.14/day; dual DGX breaks even in 24 years", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=50212065", "title": "Kimi K3.1 Pro and K3.1 Pro Highspeed", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49996437", "title": "Claude Haiku 5.5", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49672966", "title": "Ask HN: What default model do you use and why?", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49411102", "title": "Anthropic's best AI model struggles to attract users as cheaper tools thrive", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {}, "2026-01": {}, "2026-02": {"compare": 1, "warn": 1}, "2026-03": {"recommend": 2}, "2026-04": {"recommend": 3, "warn": 1, "compare": 1, "neutral": 1}, "2026-05": {}, "2026-06": {"neutral": 1, "recommend": 5}, "2026-07": {"recommend": 4, "neutral": 1}, "2026-08": {"neutral": 1, "recommend": 3}, "2026-09": {"recommend": 1}, "2026-10": {"neutral": 2, "recommend": 1}}, "quotes": [{"id": 803443955, "item_id": 50215247, "url": "https://news.ycombinator.com/item?id=50215247", "by": "ricardobeat", "date": "2026-10-23", "stance": "neutral", "quote": "then handed off to DeepSeek or Minimax", "excerpt": "I had the lowest plan ($20/month I think) and it literally gave me fifteen, maybe twenty minutes tops, using the cheaper K3-256K variant. Basically enough to fill the context about 1.2x and that was it. It's a really smart model but way too expensive. The same $20 goes 100x further with Claude. I used it for the initial task planning while the plan didn't expire, then handed off to DeepSeek or Minimax.", "context": "", "use_case": "implementation after planning", "switched_from": "", "story": "Kimi K3.1 Pro and K3.1 Pro Highspeed"}, {"id": 802557504, "item_id": 50159844, "url": "https://news.ycombinator.com/item?id=50159844", "by": "jauntywundrkind", "date": "2026-10-20", "stance": "recommend", "quote": "I really really enjoyed talking with MiniMax M3 a lot.", "excerpt": "Skepticism is due! Good! The numbers don't mean much! Yes! I personally enjoy seeing how different models view problems. Take big plans from big models, then see how other agents reinterpret a stance. See which directions they go. I love the diversity of views they produce. It's not really relevant what the numbers are. I really really enjoyed talking with MiniMax M3 a lot. And M2.7. it had a good world view, wrote t", "context": "", "use_case": "exploring problems and plans", "switched_from": "", "story": "MiniMax M3.1 Flash"}, {"id": 802557505, "item_id": 50159844, "url": "https://news.ycombinator.com/item?id=50159844", "by": "jauntywundrkind", "date": "2026-10-20", "stance": "recommend", "quote": "And M2.7. it had a good world view", "excerpt": "Skepticism is due! Good! The numbers don't mean much! Yes! I personally enjoy seeing how different models view problems. Take big plans from big models, then see how other agents reinterpret a stance. See which directions they go. I love the diversity of views they produce. It's not really relevant what the numbers are. I really really enjoyed talking with MiniMax M3 a lot. And M2.7. it had a good world view, wrote t", "context": "", "use_case": "exploring problems and plans", "switched_from": "", "story": "MiniMax M3.1 Flash"}, {"id": 800071156, "item_id": 50004447, "url": "https://news.ycombinator.com/item?id=50004447", "by": "jeremyjh", "date": "2026-10-08", "stance": "neutral", "quote": "I also have a $10 minimax plan and use M3 for exploration and library roles", "excerpt": "I use oh-my-pi (omp.sh) - mostly with stock settings and skills but its a \"fully loaded\" harness that you don't really have to add anything to. I do change a couple of things: I set it to prefer subagents, and I enable rewind. It is crazy that rewind is not a default, it saves a TON of context - if the agent goes down a crazy path that burns a lot of tokens it can rewind to an earlier checkpoint with the exploration ", "context": "", "use_case": "exploration and library roles", "switched_from": "", "story": "Claude Haiku 5.5"}, {"id": 794842179, "item_id": 49677636, "url": "https://news.ycombinator.com/item?id=49677636", "by": "ricardobeat", "date": "2026-09-12", "stance": "recommend", "quote": "My current go-tos are DS Flash, Minimax M3 for subagents.", "excerpt": "Opus 5 at low effort is quite good at most tasks, and will go a long way even on the base plan. At higher thinking levels it is unbearably cautious and verbose - it drives me mad. I don\u2019t have a use for Fable at current prices. My current go-tos are DS Flash, Minimax M3 for subagents. Adding Muse Spark 1.3 to that mix but not convinced yet.", "context": "", "use_case": "subagents", "switched_from": "", "story": "Ask HN: What default model do you use and why?"}, {"id": 790685683, "item_id": 49417855, "url": "https://news.ycombinator.com/item?id=49417855", "by": "dkersten", "date": "2026-08-24", "stance": "recommend", "quote": "a MiniMax subscription for high-token-use tasks that don\u2019t need the best model (for $20 I get what feels like infinite tokens)", "excerpt": "I stopped using Claude because of this BS. If I pay for a service, I want to know what I\u2019m paying for, I want it to be predictable. Anthropic have been anything but. Flip flopping on model availability, model access behind an opaque filter, their past behaviour of model degradation as they prepared their next model\u2026 these are not signs of a reliable service. I\u2019ve mostly settled on using a mixture of open weights mode", "context": "", "use_case": "high-token-use tasks", "switched_from": "Claude", "story": "Anthropic's best AI model struggles to attract users as cheaper tools thrive"}, {"id": 787894672, "item_id": 49243417, "url": "https://news.ycombinator.com/item?id=49243417", "by": "notjes", "date": "2026-08-10", "stance": "recommend", "quote": "I have been using MiniMax M3 and I am really surprised by the quality this model spits out.", "excerpt": "Since the announcement of DeepSeek price hike, I have been using MiniMax M3 and I am really surprised by the quality this model spits out. Subjectively speaking, the bullshit MM3M produces is waaay less than DS0731. Could it be a result of less hallucinations than DS0731? https://artificialanalysis.ai/evaluations/omniscience#aa-omniscience-hallucination-rate Maybe hallucination is good for prototyping and creative wo", "context": "", "use_case": "coding", "switched_from": "DeepSeek", "story": "DeepSeek costs OpenCode Go user $1.14/day; dual DGX breaks even in 24 years"}, {"id": 787886001, "item_id": 49242875, "url": "https://news.ycombinator.com/item?id=49242875", "by": "jauntywundrkind", "date": "2026-08-10", "stance": "recommend", "quote": "I've been very surprised with MiniMax M3, not as a particularly good architect, but at it's very good ability to state the problem elegantly", "excerpt": "That's almost half way to their stated 6X usage goal! Just from DeepSeek. > With Go, you pay $10/month and we aim to give you 6x that in usage. For most models, we make this work through bulk discounts and reserved GPU capacity. We then pass those savings on to you through the 6x multiplier. https://opencode.ai/docs/go/#why-some-models-have-lower-usage I think it's back to 2X usage, meaning it's cheaper token-burn th", "context": "", "use_case": "design work", "switched_from": "", "story": "DeepSeek costs OpenCode Go user $1.14/day; dual DGX breaks even in 24 years"}, {"id": 787460866, "item_id": 49216304, "url": "https://news.ycombinator.com/item?id=49216304", "by": "naught0", "date": "2026-08-07", "stance": "neutral", "quote": "stepping up to minimax m3 if that doesn't cut it", "excerpt": "I enjoy using opencode go to play around with a lot of different models. I wind up using deepseek v4 flash for most everything, stepping up to minimax m3 if that doesn't cut it, finally preferring GLM for complex tasks or important planning I want to go right the first time I recommend opencode or something akin to it to play with models. Any big model updates or hot new ones will naturally run across your desk that ", "context": "", "use_case": "harder coding tasks", "switched_from": "", "story": "DeepSeek V4 Flash 0731"}, {"id": 785312919, "item_id": 49082057, "url": "https://news.ycombinator.com/item?id=49082057", "by": "Frannky", "date": "2026-07-28", "stance": "recommend", "quote": "MiniMax M3 for agents", "excerpt": "I think they're doing something wrong. I have access to better models, but I just use 4.6 and 4.8. Fable and Opus 5 burn tokens too fast for me. I also switched to a cheaper plan to work less and to experiment with OpenCode and open models when I hit the 5h limits. I'm doing this partly because I feel they've already started the enshittification of the old models, and I want to be able to just stop paying for the sub", "context": "", "use_case": "agents", "switched_from": "", "story": "\"Opus 5 is a really bad model\""}, {"id": 783809105, "item_id": 48988069, "url": "https://news.ycombinator.com/item?id=48988069", "by": "ricardobeat", "date": "2026-07-20", "stance": "recommend", "quote": "switched to Minimax M3 continuing from the same context", "excerpt": "I just did this five minutes ago. Started a task with Kimi K3, noticed cost going up, switched to Minimax M3 continuing from the same context. Really easy to do with Crush/OpenCode. It works quite well most of the time. For very deeply technical tasks, where exploration involves multiple agents (and blowing up context limits a few times), I'll have a plan written down to disk that will be much longer than 2k tokens.", "context": "", "use_case": "cheaper coding continuation", "switched_from": "Kimi K3", "story": "You only need the frontier model for one single edit"}, {"id": 783560962, "item_id": 48972560, "url": "https://news.ycombinator.com/item?id=48972560", "by": "p1necone", "date": "2026-07-19", "stance": "neutral", "quote": "Minimax M3 feels okay but definitely a step down from GLM-5.2.", "excerpt": "I use GLM-5.2 as an orchestrator model which delegates to Deepseek subagents (v4 flash or pro depending on complexity) and it works pretty well for a quite complex compiler codebase when I do deep enough up front planning for features/fixes. If you believe the benchmarks Deepseek v4 is pretty shitty at long running work in large codebases, but really really good at self contained algorithmic/math reasoning - which is", "context": "", "use_case": "fallback orchestrator", "switched_from": "", "story": "Qwen 3.8 Max Preview"}, {"id": 783385042, "item_id": 48961565, "url": "https://news.ycombinator.com/item?id=48961565", "by": "ignoramous", "date": "2026-07-18", "stance": "recommend", "quote": "subscribing (directly [0]) with providers that are transparent like MiniMax, DeepSeek, Xiaomi, & Z.ai.", "excerpt": "> tried Kimi K3 on a task I've done with every other model I use regularly and found it chewed a lot longer on the problem and ate up almost the entirety of a 5 hour usage limit on their $19 plan ArtificialAnalysis puts Kimi K3 just below DeepSeek v4 & GLM 5.2 in token use per task, which is about 2x to 3x more tokens than Grok 4.5: https://x.com/ArtificialAnlys/status/2077832879187620192 > Subscription usage limits ", "context": "", "use_case": "coding subscription", "switched_from": "", "story": "The Kimi K3 Moment"}, {"id": 781066306, "item_id": 48816644, "url": "https://news.ycombinator.com/item?id=48816644", "by": "jeremyjh", "date": "2026-07-07", "stance": "recommend", "quote": "I configured it to use MiniMax M3 for this task", "excerpt": "If you use a good harness or add the right tools and plugins both the image and web search issues mentioned are non-issues. oh-my-pi (omp.sh) handles images for text models out of the box - as long as you have any vision capable provider enabled, it will be used when you paste images to a text model. Rather than let it guess I configured it to use MiniMax M3 for this task (as well as other utility tasks like code exp", "context": "", "use_case": "image handling and utility tasks", "switched_from": "", "story": "GLM 5.2 and the coming AI margin collapse"}, {"id": 779827250, "item_id": 48739203, "url": "https://news.ycombinator.com/item?id=48739203", "by": "Imustaskforhelp", "date": "2026-06-30", "stance": "recommend", "quote": "Z.Ai subscription/ Kimi subscription / MiniMax subscriptions are good", "excerpt": "I think that there are some subscriptions to go by. Z.ai subscription might still be interesting. I once haggled with kimi to get it for 1$ per month. I can only help in providing pointers: If you wish to go Non-API but rather subscription route: Z.Ai subscription/ Kimi subscription / MiniMax subscriptions are good. You could also take a look at ollama subscription and opencode subscriptions. If you wish to go API ro", "context": "", "use_case": "cheap coding model subscription", "switched_from": "", "story": "Claude Code Is Steganographically Marking Requests"}, {"id": 779437793, "item_id": 48714862, "url": "https://news.ycombinator.com/item?id=48714862", "by": "uluckydev", "date": "2026-06-28", "stance": "recommend", "quote": "When M3 came out, I started using it, and it was actually really good.", "excerpt": "I used Claude a lot, but with Claude Code it takes a lot of context window, and it's very pricey, to be honest. Then I shifted towards Minimax. I used the coding plan because it's cheaper, but it still gets the job done. When M3 came out, I started using it, and it was actually really good. After that, I shifted towards OpenCode for my AI agent, and that's been really good as well. The best thing I realized is that i", "context": "", "use_case": "AI coding", "switched_from": "Claude Code", "story": "Semgrep: GLM 5.2 beats Claude in our Cyber Benchmarks"}, {"id": 778082049, "item_id": 48630128, "url": "https://news.ycombinator.com/item?id=48630128", "by": "jeremyjh", "date": "2026-06-22", "stance": "recommend", "quote": "I also use MiniMax-M3 in utility roles like explore/library tasks.", "excerpt": "This is the way but Anthropic doesn\u2019t make it easy, so I use GPT 5.5 in that role since I can use my subscription in OpenCode or OMP. I also use MiniMax-M3 in utility roles like explore/library tasks. I\u2019ve had a z.ai subscription for several months so I\u2019m on the older pricing. I\u2019m really not sure it would make sense to do this at current rates - I could bump my Codex plan instead.", "context": "", "use_case": "explore/library tasks", "switched_from": "", "story": "GLM 5.2 vs. Opus"}, {"id": 777904162, "item_id": 48619010, "url": "https://news.ycombinator.com/item?id=48619010", "by": "pet_the_bird", "date": "2026-06-21", "stance": "recommend", "quote": "it felt like such a breath of fresh air when I used the reference full-size model", "excerpt": "Highy anecdotal: I have tried various self-hosted models using both vllm and llama.cpp. I am in a situation where I have access to large amount of memory (~320 GB). While experimenting with quantization I found that there is a non-trivial tradeoff between quality and memory footprint. Overall my experience follows the reported pattern of \"2-bit is mwah, 4-bit half decent and 6-bit required for programming. Still, alt", "context": "has access to ~320 GB memory", "use_case": "programming", "switched_from": "", "story": "Two Qwen3 models on one DGX Spark: the residency math"}, {"id": 777714768, "item_id": 48607173, "url": "https://news.ycombinator.com/item?id=48607173", "by": "throwa356262", "date": "2026-06-20", "stance": "recommend", "quote": "Currently testing M3 for agentic tasks. It works OK and their token plan is very cheap. Highly recommend for claw / hermes type of work.", "excerpt": "Since I quit my Claude subscription, every month I spend $20 (the cost of CC pro plan) playing around with new models and new providers. Currently testing M3 for agentic tasks. It works OK and their token plan is very cheap. Highly recommend for claw / hermes type of work. Tested GLM 5.1 for coding last month and it burned through my tokens a bit too quickly, but it worked well enough. Don't seem to have a coding pla", "context": "", "use_case": "agentic tasks", "switched_from": "Claude", "story": "MiniMax M3 vs. GLM 5.2: Codegen comparison across autonomous coding tasks"}, {"id": 776704400, "item_id": 48544025, "url": "https://news.ycombinator.com/item?id=48544025", "by": "_davide_", "date": "2026-06-15", "stance": "neutral", "quote": "i used to mix remote and local minimax 2.7(q3) on my strix halo", "excerpt": "i used to mix remote and local minimax 2.7(q3) on my strix halo, it run at 30 tg and 220 tokens pp... it was a bit painful slow, but it was a good feeling i could stay offline. unfortunately m3 which is at opus .8 levels is 460b parameters and doesn't even fit in 128gb of memory, let alone a big context. strix halo feels like a toy for ai purposes. https://kyuz0.github.io/amd-strix-halo-toolboxes/", "context": "", "use_case": "local coding model", "switched_from": "", "story": "Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?"}, {"id": 767437764, "item_id": 47964860, "url": "https://news.ycombinator.com/item?id=47964860", "by": "alexjplant", "date": "2026-04-30", "stance": "recommend", "quote": "pay as you go on cheaper models like MiniMax you can get a lot done for not a lot of coin", "excerpt": "> casting about for viable egress paths if I need to flee Claude Code Check out OpenCode (the OSS product [1]) and OpenCode Go/Zen (the LLMaaS [2]). Use a more expensive model with larger context (like GLM-5.1) for orchestration and cheaper models for coding and iteration on acceptance criteria (writing and passing tests). I also throw a more expensive vision-capable model into the mix like Gemini 3 Flash to iterate ", "context": "", "use_case": "cheap coding iteration", "switched_from": "", "story": "Claude Code refuses requests or charges extra if your commits mention \"OpenClaw\""}, {"id": 766544707, "item_id": 47909044, "url": "https://news.ycombinator.com/item?id=47909044", "by": "DeathArrow", "date": "2026-04-26", "stance": "recommend", "quote": "I used Claude Code with GLM 5.1, MiniMax M2.7, Kimi K2.6 and had pretty good results.", "excerpt": ">I initially drafted this before my last post on how Claude Code is getting worse. I'm putting it out now so I can reference it in a future post on OpenCode. IMO, what is getting worse is not Claude Code, the CLI tool, but Anthropic API. That's what most people experience. I used Claude Code with GLM 5.1, MiniMax M2.7, Kimi K2.6 and had pretty good results. I prefer Claude Code over OpenCode because most plugins and ", "context": "", "use_case": "coding with Claude Code", "switched_from": "", "story": "It's OK to Use Agentic to Revive the Projects You Never Were Going to Finish"}, {"id": 766280707, "item_id": 47892544, "url": "https://news.ycombinator.com/item?id=47892544", "by": "DeathArrow", "date": "2026-04-24", "stance": "neutral", "quote": "I use Claude Code with GLM, Kimi and MiniMax models.", "excerpt": "I use Claude Code with GLM, Kimi and MiniMax models. :) I was worried about Anthropic models quality varying and about Anthropic jacking up prices. I don't think Claude Code is the best agent orchestrator and harness in existence but it's most widely supported by plugins and skills.", "context": "", "use_case": "model behind Claude Code", "switched_from": "", "story": "I Cancelled Claude: Token Issues, Declining Quality, and Poor Support"}, {"id": 764018611, "item_id": 47751163, "url": "https://news.ycombinator.com/item?id=47751163", "by": "Robelkidin", "date": "2026-04-13", "stance": "recommend", "quote": "MiniMAX and Claude models tend to be better AI models at doing agentic tasks.", "excerpt": "I have been experimenting with openclaw for the past 3 months and I am Software Engineer and I had difficulties setting it up and managing it. I almost bought Mac Studio to run it, bought into the hype. I do believe it absolutely amazing product. I recently launched AlitaGPT.COM (Alita.com was already registered) to provide a managed openclaw with hundreds of custom application integration. I used GCP GKE -Autopilot ", "context": "software engineer", "use_case": "agentic tasks", "switched_from": "", "story": "OpenAI GPT worst AI GPT/model vs. Claude/MiniAAX"}, {"id": 763837187, "item_id": 47739824, "url": "https://news.ycombinator.com/item?id=47739824", "by": "rzkyif", "date": "2026-04-12", "stance": "compare", "quote": "It's quite cheap and seems to be better than Gemini 3 Flash", "excerpt": "My personal experience is way different: I struggle to burn through more than 50% of the 5 hour limit For context, with Google AI Pro, I can burn through the Antigravity weekly limit in 1-2 hours if I force it to use Gemini 3.1 Pro. Meanwhile Gemini 3 Flash is basically unlimited but frequently produces buggy code or fail to implement things how I personally would (felt like it doesn't \"think\" like a software dev) I ", "context": "", "use_case": "coding via Cline", "switched_from": "", "story": "Pro Max 5x Quota Exhausted in 1.5 Hours Despite Moderate Usage"}, {"id": 763814912, "item_id": 47738432, "url": "https://news.ycombinator.com/item?id=47738432", "by": "fg137", "date": "2026-04-12", "stance": "warn", "quote": "the quality is really far from Claude models.", "excerpt": "What's people's experience of using MiniMax for coding? I had a really bad time with it. I use (real) Claude Code for work so I know what a good model feels like. MiniMax's token plan is nice but the quality is really far from Claude models. I needed to constantly \"remind\" it to get things done. Even for a four sentence prompt in a session that is well below the context window, MiniMax would ignore half of it. This h", "context": "uses Claude Code for work", "use_case": "coding", "switched_from": "", "story": "MiniMax M2.7 Is Now Open Source"}, {"id": 763550688, "item_id": 47721918, "url": "https://news.ycombinator.com/item?id=47721918", "by": "cat_plus_plus", "date": "2026-04-10", "stance": "recommend", "quote": "I just got MiniMax $200/year token plan. Usually it works fine for daily coding", "excerpt": "I just got MiniMax $200/year token plan. Usually it works fine for daily coding, if it gets stuck I pay for some Claude API calls through Roo gateway. Unlike other plans, this one officially supports running OpenClaw or other API workflows and doesn't suspend you long term if you use too many tokens, just set rate per few hours.", "context": "", "use_case": "daily coding", "switched_from": "", "story": "Reallocating $100/Month Claude Code Spend to Zed and OpenRouter"}, {"id": 758058291, "item_id": 47378643, "url": "https://news.ycombinator.com/item?id=47378643", "by": "Cytoplast3528", "date": "2026-03-14", "stance": "recommend", "quote": "I think only Claude Sonnet/Opus, GPT 5.2+, Minimax M2.5 are useful.", "excerpt": "I think only Claude Sonnet/Opus, GPT 5.2+, Minimax M2.5 are useful. They are all nearly impossible to self-host, unfortunately.", "context": "", "use_case": "", "switched_from": "", "story": "Ask HN: How do you use local LLMs productively?"}, {"id": 757094944, "item_id": 47318434, "url": "https://news.ycombinator.com/item?id=47318434", "by": "radio879", "date": "2026-03-09", "stance": "recommend", "quote": "Current best models for that - MinimaxM2.5, Gemini 3.1 Flash", "excerpt": "With all of the talk about \"LLMs are non deterministic\" I am trying to build my own tool (at this exact moment, since I seem to have gotten really good at this) which does some of what these OpenClaw/Clawdbots do for people, but instead of having these LLMs/agents handle everything I am trying to do as much of it as pure deterministic code as possible. I test things once, like daily automations, and they will repeat ", "context": "works with LLMs daily for 8-15 hours", "use_case": "agentic tool use", "switched_from": "", "story": "Ask HN: Read\u2011only LLM tool for email triage and knowledge extraction?"}, {"id": 753243153, "item_id": 47077697, "url": "https://news.ycombinator.com/item?id=47077697", "by": "sosodev", "date": "2026-02-19", "stance": "warn", "quote": "I suspected that MiniMax M2.5 might be a bit too much for this board.", "excerpt": "Thanks for the additional info. I suspected that MiniMax M2.5 might be a bit too much for this board. 230B-A10B is just a lot to ask of the 395+ even with aggressive quantization. Particularly when you consider that the model is going to spend a lot of tokens thinking and that will eat into the comparatively smaller context window. I switched from the Unsloth 4-bit quant of Qwen3 Coder Next to the official 4-bit quan", "context": "running local models on a 395+ board", "use_case": "local agentic coding", "switched_from": "", "story": "Why AI Velocity Is Becoming a Debt Accelerator"}, {"id": 751910689, "item_id": 46994418, "url": "https://news.ycombinator.com/item?id=46994418", "by": "mchusma", "date": "2026-02-12", "stance": "compare", "quote": "I do believe if there is something that comes up that is literally continuous, would be interesting, but I'm not sure about it right now.", "excerpt": "This is cool, but they mentioned affordability, and said this is about $1/hour to run, which is about what I pay for claude code on $200/mo plan. This is not literally true, sometimes I'm running up to 3 concurrent intermittently throughout the day for maybe 60 hours per week. So I do believe if there is something that comes up that is literally continuous, would be interesting, but I'm not sure about it right now. I", "context": "", "use_case": "", "switched_from": "", "story": "MiniMax M2.5 released: 80.2% in SWE-bench Verified"}]}