StepFun 3.5 Flash is better compared to google's gemini 3 flash which is surprisingly good and pretty costly, and to GLM-5.
I find this outcome ironic given minimax's more aggressive marketing and large-scale distillation accusations from Anthropic specifically accusing minimax but not StepFun.
I can only wonder about the true underlying reasons, but deducing from public information I suspect that minimax simply has weaker, benchmaxx-targeting post-training R&D and leans more on distillation of western frontier models, while StepFun has extensive post-training with lots of hard-won custom R&D and internal large-scale distillation teachers.
I tried it out a bunch and it seems good. I can't really tell if it's better or worse than most of these other models in such a short time though.
Where GLM 5 is strictly worse for me though, compared to StepFun, is long-form content generation (planning, research documents) - but this can be said about geminis too and these are obviously very smart models.
Given the free option I'd explore GLM 5 more, but if I had to pay for it myself ofc I'd choose stepfun every time. Basically I think right now the optimal configuration for maximizing output of correct software features per dollar involves using StepFun or its future class competitor for bulk coding and first stage code review.
Maybe I need to write a blogpost about it after all.
What's really surprising to me is the cost of the model. It's definitely very good for its price. DeepSeek is the only one that offers and competition to it at that price point (GLM 5 is literally 10x more expensive).
It’s 2× faster than its competitors. For tasks where “one-shotting” is unrealistic, a fast iteration loop makes a measurable difference in productivity.
Even purely pragmatically, StepFun covers 95% of my research+SWE coding needs, and for the remaining 5% I can access the large frontier models. I was surprised StepFun is even decent at planning and research, so it is possible to get by with it and nothing else (1), but ofc for minmaxing the best frontier model is still the best planner (although the latest deepseek is surprisingly good too).
Finally we are at a point where there is a clear separation of labor between frontier & strong+fast models, but tbh shoehorning StepFun into this "strong+fast" category feels limiting, I think it has greater potential.
Claude code always give me rate limits. Claude through copilot is a bit slow, but copilot has constant network request issues or something, but at least I don't get rate limited as often.
At least local models always work, is faster (50+ tps with qwen3.5 35b a4b on a 4090) and most importantly never hit a rate limit.
> 50+ tps with qwen3.5 35b a4b on a 4090
But qwen3.5 35b is worse than even Claude Haiku 4.5. You could switch your Claude Code to use Haiku and never hit rate limits. Also gets similar 50tps.
My goto proprietary model in copilot for general tasks is gemini 3 flash which is priced the same as haiku.
The qwen model is in my experience close to gemini 3 flash, but gemini flash is still better.
Maybe it's somewhat related to what we're using them for. In my case I'm mostly using llms to code Lua. One case is a typed luajit language and the other is a 3d luajit framework written entirely in luajit.
I forgot exactly how many tps i get with qwen, but with glm 4.7 flash which is really good (to be local) gets me 120tps and a 120k context.
Don't get me wrong, proprietary models are superior, but local models are getting really good AND useful for a lot of real work.
To be clear I never said they weren’t strong or useful. I use them for some small tasks too.
I said they’re not equivalent to SOTA models from 6 months ago, which is what is always claimed.
Then it turns into a Motte and Bailey game where that argument is replaced with the simpler argument that they’re useful for open weights models. I’m not disagreeing with that part. I’m disagree with the first assertion that they’re equivalent to Sonnet 4.5
Maybe my detailed, requirement-based/spec-based prompting style makes the difference between anthropic's and OSS models smaller and people just like how good Anthropic's models are at reading the programmer's intent from short concise prompts.
Frankly, I think the 1:1 equivalent is an impossible standard given the set of priorities and decisions frontier labs make when setting up their pre-, mid- and post-training pipelines, and benchmark-wise it is achievable for a smaller OSS model to align with Sonnet 4.5 even on hard benchmarks.
Given the relatively underwhelming Sonnet 4.5 benchmarks [1], I think StepFun might have an edge over it esp. in Math/STEM [2] - even an old deepseek-3.2 (not speciale!) had a similar aggregate score. With 4.6 Anthropic ofc vastly improved their benchmark game, and it now truly looks like a frontier model.
1. https://artificialanalysis.ai/models/claude-4-5-sonnet-think... 2. https://matharena.ai/models/stepfun_3_5_flash
I'm working on a pretty complex Rust codebase right now, with hundreds of integration tests and nontrivial concurrency, and stepfun powers through.
I have no relation to stepfun, and I'm saying this purely from deep respect to the team that managed to pack this performance in 196B/11B active envelope.