On 2026-10-15 StepFun released open weights for Step 5 Preview as promised: a 600B-total / 27B-active sparse MoE with up to 1M-token context and vision input, aimed at agentic software engineering and knowledge work. Checkpoints are on Hugging Face (FP8, BF16, NVFP4) and ModelScope. Official scores include StepCodeBench 49.0%, FrontierFinance 66.4%, and Artificial Analysis Intelligence Index 44. September coverage was API/index only; this is the first public weight drop.
Key Takeaways
- ✓Specs: sparse MoE 600B total / 27B active, up to 1M-token context, text+image input (model page)
- ✓Weights:
stepfun-ai/Step-5-Previewplus BF16 and NVFP4 on Hugging Face and ModelScope (HF card) - ✓Coding: StepCodeBench avg@4 49.0% (leads the open-weight set StepFun cites); DeepSWE v1.1 67.7% (official table)
- ✓Finance / index: FrontierFinance 66.4%; Artificial Analysis Intelligence Index 44 (matches earlier AA coverage)
- ✓Serve via official vLLM / SGLang Docker images and OpenAI-compatible APIs; license is StepFun Model License 1.0 (Apache-2.0-style with commercial conditions)
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
StepFun released open weights for Step 5 Preview on 2026-10-15, fulfilling the October 15 open-weight plan noted in earlier Artificial Analysis coverage. The model is a sparse MoE with about 600B total parameters and 27B active per token, up to 1M tokens of context, and text plus image input. It is positioned for agentic software engineering, long-horizon tool use, and professional knowledge work with a finance focus.
Official numbers from the StepFun model page and Hugging Face card (not extrapolated): StepCodeBench (High) 49.0%, DeepSWE v1.1 67.7%, ProgramBench 80.5%, FrontierFinance 66.4%, and Artificial Analysis Intelligence Index 44. Gaps remain versus closed frontier models such as Claude Opus 5 and GPT-6 Astra on several coding suites, but among the open-weight peers StepFun cites it leads on StepCodeBench.
Weights are on Hugging Face as stepfun-ai/Step-5-Preview plus BF16 and NVFP4 variants, and on ModelScope. Serve with the published vLLM / SGLang Docker images and OpenAI-compatible APIs; recommended sampling is temperature=1.0, top_p=0.95. License is StepFun Model License 1.0 (Apache-2.0-style with attribution and commercial conditions). The news today is downloadable weights, not a new API price.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.