StepFun promotes Step 5 Preview with benchmark claims and API documentation
Overview
StepFun is promoting its Step 5 Preview model, directing readers to a product page with an overview, benchmarks and demos, and a platform page with technical specifications and API documentation.
The company's own promotion is the source of this announcement.
OpenRouter, a model-access platform, reports that StepFun's eight-benchmark evaluation shows Step 5 Preview beating Kimi K3 and GLM-5.3 on six benchmarks. Cited scores include DeepSWE at 67.7, ProgramBench at 80.5, StepCodeBench at 49.0 and FrontierFinance at 66.4. These comparative results come from StepFun's evaluation as relayed by OpenRouter, and have not been independently verified.
Written by AI from the articles below · updated Oct 8, 9:02 PM ET
Check the sources:
Developments
2 developments
- Oct 8, 9:33 AM ET · 1 articleOpenRouter reports Step 5 Preview beats Kimi K3 and GLM-5.3 on 6 of 8 StepFun benchmarksOpenRouter: Across StepFun's 8-benchmark eval, Step 5 Preview beats Kimi K3 and GLM-5.3 on 6, including DeepSWE (67.7), ProgramBench (80.5)…
- Oct 8, 9:08 AM ET · 1 articleStepFun promotes Step 5 Preview model overview, benchmarks, and API docsStepFun: Want to learn more about Step 5 Preview? Model overview, benchmarks & demos ↓ Technical specs & API docs ↓
Article timeline
Follow the coverage from different perspectives. Times are ET.
- OpenRouterAcross StepFun's 8-benchmark eval, Step 5 Preview beats Kimi K3 and GLM-5.3 on 6, including DeepSWE (67.7), ProgramBench (80.5)…
AI…StepCodeBench (49.0), and FrontierFinance (66.4).
Heat trend
Not enough continuous observations to show a trend yet.