StepFun's Step 5 Preview Surfaces: 600B Params, Output Priced at 1/5 of Kimi K3
teortaxesTex · x · 2026-09-20
StepFun hasn't officially announced Step 5, but Artificial Analysis has benchmarked Step 5 Preview via the StepFun API. Per AA, it's a multimodal reasoning model with about 600B total parameters and 27B activated per inference, supporting image input and a 1M-token context.
Scores: 44 on the latest Intelligence Index, tied with Kimi K3 Max; 33% on Terminal-Bench 4.0 (Kimi K3: 13%); both at 59% on SciCode. Pricing: $1 per million input tokens and $2.7 per million output tokens (Kimi K3: $3 / $15) — output costs under one-fifth of K3, and AA's full Intelligence Index run cost about a quarter of K3's.
Related event: StepFun Launches Step 5 Preview: 600B MoE Flagship, Open-Weights on Oct 15(17 posts)→
More from Models
- a16z's Martin Casado: Google's Carbon is 'really good at code' — xennygrimmato_ · 2026-11-12
- OpenRouter launches Extractions API: typed JSON field extraction from text, images and PDFs — michellechen · 2026-11-12
- Google launches Gemini Decisions API with Decision 1 model: closed-set choices, free output tokens — tokumin · 2026-11-12
- Google's busy fall: Gemini 4 Argon launches, plus OpenAI models on Google Cloud — samcharrington · 2026-11-12
- Google Cloud Now Hosts Gemini, Claude, and GPT — a Hyperscaler First — samcharrington · 2026-11-12
- Carbon + /swarm in Antigravity: Big Model Thinks, Lighter Models Work — samcharrington · 2026-11-12