3 ms·
Ask HN: Which cheap Chinese LLM are you using?
In the last one or two months, starting from DeepSeek V4 Pro, there are quite many low-price Chinese models coming out. Their performance looks more or less similar to me: Mimo V2.5 Pro, MiniMax M3, and the just released GLM 5.2, etc.
Which model are you using now? Why? What are the good and bad parts?
- verdverm 4mo agoDeepSeek v4 let's me do more with my quota cheap, fast, not as good, but more gets done without giving Big Ai any more money
- taris2 4mo ago[dead]
- sermakarevich 4mo agokimi k2.7 code looks great on this ranking https://x.com/prz_chojecki/status/2065741640635990128?s=20 https://x.com/prz_chojecki/status/2065741640635990128?s=20 I use qwen3.6:36B locally
- zionsati 4mo agodeepseek v4 pro - but honestly it is comparable to gpt-5.4-mini, far from GPT5.4 let alone GPT5.5! Its advantage is really just its pricing. I'm going to give Kimi K2.7 a try - with K2.6, even its cloud chat locks up all the time, so it really didn't give me much confidence at all for agentic coding.
- linzhangrun 4mo agoI think it is more like a post-training issue. DeepSeek is cheap enough for heavy token use, so it fits Hermes very well.
- zionsati 4mo agoI did a bit of digging after this to see if my feeling was accurate, and lo-and-behold: https://deepswe.net/ https://deepswe.net/ Look at the gap between gpt5.4-mini vs deepseek v4 pro!
- cyanydeez 4mo agoqwen3.6 35b. other peoples compute resources are unreliable ar best.
- greenoracle9 4mo agoThe main reason I'm using Chinese LLM is cost. Minimax M3 is a good deal with large context window. But, M3 jumps into implementation too quickly even when a task is clearly defined. It misses tests or edge cases, and occasionally lose track during longer work.
- jeffyaw 4mo agoi just use review skills on my sessions/prs extensively until they're in a good place. then coverage skills (which add test coverage). i built in skills to work with M3 in a service called typed, an ai cli. it uses m3 under the hood (up to ~500k tokens), then switches to deepseek for up to 1M. a few bells and whistles added of typescript/python coding optimization. and just built a custom TUI frontend for it (initially works with the claude code tui and still does). to toggle the typed tui you can run: typed cli on typed cli off
- rubslopes 4mo agoI was going back and forth between kimi k2.6 and deepseek v4 pro, but it's been 2 weeks that I've been using deepseek alone. It's very good and very cheap, I pay the opencode go $10 subscription and don't have to worry about quotas.