As of Oct 15, DeepInfra now serves StepFun's Step 5 Preview, a 600B open-weight MoE with 27B active parameters, 1M-token context and text plus vision input, in FP8.
@DeepInfra on X published this on Oct 15, 2026 (source: x.com).
DeepInfra now serves StepFun's Step 5 Preview, a 600B open-weight MoE with 27B active parameters, 1M-token context and text plus vision input, in FP8. Pricing is $1.00 per million input tokens, $2.70 per million output tokens and $0.05 per million cached tokens, and StepFun welcomed the launch.
Hugging Face @stepfun-ai published this on Oct 14, 2026 (source: huggingface.co).
StepFun published stepfun-ai/Step-5-Preview-NVFP4 on Hugging Face, an 8-bit ModelOpt NVFP4 quantized build of Step 5 Preview, its flagship sparse mixture-of-experts model for agentic work with 600B total and 27B active parameters and up to 1M context. The card points to the FP8 and BF16 releases of the same model and carries StepFun's model license. Tags list safetensors, step5_vl and modelopt.
@StepFun_ai on X published this on Oct 9, 2026 (source: x.com).
StepFun says its Step 5 Preview model is free to use in Kilo Code for one week.
Items older than 72 hours never go in Top stories and show their original date.