Model API layer for coding agents providing fast general models and specialized submodels for code editing, search, compaction, and UI testing.
Morph is a specialized inference API for coding agents that offers fast general models for the primary agent loop plus small task-specific submodels: Fast Apply (code edits at 10,500 tok/s), WarpGrep (code search, ranked #1 on SWE-Bench Pro), Compact (context compaction at 33,000 tok/s), Reflexes (semantic classifiers for traces), and Glance (AI browser and mobile testing). All models run through one OpenAI-compatible API on custom GPU kernels. Pricing is usage-based per million tokens with no per-seat fees; a free tier offers 200 requests/month. Morph is a product of Morph.
For people
For agents