Step 3.5 Flash

At a glance

What Step 3.5 Flash does

Tool use, Streaming, Structured output, Sparse MoE (196B / 11B active), Fast inference (197 tok/s), Function calling, Self-hostable

Best for

High-throughput low-cost reasoning, real-time agent loops, budget-tier production deployments, fastest open-weight reasoning model in its price class

Works well with

Conflicts & caveats

• Trae's built-in model menu may not include Step 3.5 Flash — Trae primarily ships with Claude/GPT/Gemini/DeepSeek.
• Google Antigravity is built around Gemini 3 Pro; Claude and recent GPT-5 models also work, but Step 3.5 Flash is not in the supported model list.
• Kiro's spec mode is tuned for Claude models — Step 3.5 Flash may work but spec-driven flows are best with Claude.
• Cursor's built-in model menu may not include Step 3.5 Flash — verify it's available in your Cursor model picker.
• OpenAI Codex (VS Code) is tied to OpenAI's GPT-5.1 Codex family — Step 3.5 Flash is not selectable inside this integration.
• Gemini CLI is locked to Google Gemini models — your LLM is Step 3.5 Flash.
• Privacy conflict: Self-hosted Step 3.5 Flash sends code to cloud @codebase (Cursor). Use local context (Continue indexing, ChromaDB, LanceDB, pgvector, Vespa self-hosted) for true privacy.
• Claude Code only works with Claude models — your LLM is Step 3.5 Flash.
• Devin uses Cognition's proprietary model internally — your selected Step 3.5 Flash will be ignored.
• OpenAI Codex (Cloud) runs GPT-5.1 Codex internally — your selected Step 3.5 Flash will be ignored.
• Jules runs Gemini models internally — your selected Step 3.5 Flash will be ignored.
• The Claude Code SDK is tuned for Claude models — Step 3.5 Flash may work via the model adapter but loses Claude-specific features (thinking, prompt caching, sub-agents).
• Lovable uses a fixed internal model — your selected Step 3.5 Flash will be ignored.
• Replit Agent 3 uses its own model selection internally — your selected Step 3.5 Flash will be ignored.