Step 5 Preview (StepFun): 600B Agent Model at $1 per Million
StepFun's flagship sparse MoE built for agentic work: 600B total parameters, 27B active per token, 1M-token context, text plus image input; API live since Sept 20, open weights planned Oct 15.
Why it matters
Frontier-class Artificial Analysis Intelligence Index score of 44 at $1/M input tokens, roughly one-eighth the per-task cost of Claude Opus 5, with strong agentic coding results (67.7% DeepSWE v1.1).
What you could build with it
A team with heavy reasoning workloads can route bulk calls to Step 5's cheap API and benchmark the output against its current model before switching.
Does it hold up?
Partly evidenced: real API benchmarks exist (Artificial Analysis: 99.8 tok/s, Intelligence Index 44), so cost-sensitive agentic coding use is credible; open weights land Oct 15, which will unlock local testing.
Built with Step 5 Preview (StepFun)
- StepFun 5 Preview on OpenVibeEval: 82/100 across 8 frontend coding runsbenchmark arena · Third-party arena scores Step 5 Preview on frontend coding tasks (8 runs, avg 82/100 accessibility).
- Step 5 Preview vs GLM-5.3: one index point, one licenceblog · Head-to-head of Step 5 Preview (AA Index 44) vs GLM-5.3 (45), flagging that weights ship Oct 15 under an unannounced licence.
- StepFun's 600B Step 5 targets AI coding agents with 1M-token contextblog · Independent launch breakdown: MoE specs, 1M context, StepCodeBench agentic coding angle.
Learn more
First spotted on article: source.
More AI models
ElevenLabs Eleven v4 + v4 Turbo: new emotive TTS architecture with a 100ms real-time variantElevenLabs released Eleven v4 and Eleven v4 Turbo on September 28: a new text-to-speech architecture with inline…models · JEV 0.72VoiceStudio: open-source, fully-local ElevenLabs alternative with voice cloning, dubbing and dictation in 646 languagesVoiceStudio is an open-source, fully-local voice AI studio: voice cloning from a 3-second sample, voice design, video…models · JEV 0.71DeepSeek V4-Flash official API: public beta with upgraded agent capabilities, Responses API and Codex supportDeepSeek's official V4-Flash is now a public beta API with massively upgraded agent capabilities - benchmark scores…models · JEV 0.69Mistral Forge: enterprise platform for training and continuously improving proprietary modelsMistral AI launched Forge (Sep 28) — a full-lifecycle model training platform (pre-training, SFT, DPO/ODPO, RL…models · JEV 0.69kev: open, trainable Jev-like family of small decision models on Qwen3.5/3.8A family of small decision models (0.8B to 27B) built on Qwen3.5/3.8 that reproduces Jev's prefill-only typed-decision…models · JEV 0.68GPT-Synopsys: OpenAI and Synopsys multi-year deal to build an AI model that operates EDA toolsOpenAI and Synopsys signed a multi-year agreement on Sep 30 to jointly develop GPT-Synopsys, a specialized model that…models · JEV 0.66
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.