Model techniques map
PublishersStepFun

StepFun

4 documents read from this publisher.

stepfun-ai/Step-3.7-Flash · Hugging Face

first party release · model card · 2026-05-30 · 16 techniques

Mixture of ExpertsMulti-Token PredictionFP8 KV-cache quantizationFP8Expert ParallelismEAGLEAutomatic tool choiceMulti-layer EAGLEReasoning parserCascade attentionFlashInfer TensorRT-LLM backendGeneric tool-call parserTensorRT-LLM multi-head attention backendFlashAttention 4llama.cppNVFP4 quantization with modelopt

Overview - StepFun Documentation

first party release · vendor docs · 2026-05-29 · 1 technique

Mixture of Experts

Step 3.7 Flash

first party release · official blog · 2026-05-09 · 4 techniques

Advisor strategyPython toolVisual Search ToolGUI operation

Step 3.7 Flash

first party release · code repo · 28 techniques

Mixture of ExpertsMulti-Token PredictionConfigurable reasoning effortSpeculative decodingFP8 KV-cache quantizationExpert ParallelismNVFP4EAGLETensor ParallelismAutomatic tool choiceAsynchronous schedulingFlashInferMulti-layer EAGLEReasoning parserCascade attentionFlashInfer TensorRT-LLM backendGGUFTensorRT-LLM multi-head attention backend256K context windowFP16 multimodal projectorIQ4_XS quantizationLanguage backboneModelOpt FP4 quantizationNVIDIA ModelOpt quantizationQ3_K_L quantizationQ4_K_S quantizationStep3p5 tool-call parserVision encoder