B2B
StartupMorph
GLM-5.3 with 1M context on inference tuned for coding agents. Reflex classifiers that catch the failures traces miss. Plus the agent toolkit: search (WarpGrep), edits (Fast Apply), context (Compact). One OpenAI-compatible API.
Milestones
Milestone
YC Summer 202317 upvotes
WarpGrep: A 20x faster subagent to grep for code
50% faster than Cognition's SWE-grep - improve Claude Code or Codex via MCP
Milestone
YC Summer 20234 upvotes
WarpGrep v2: Code Search Subagent -> #1 on SWE-Bench Pro
Code search limits coding capabiltiies. Moving it out to a specialized model wins.
Milestone
YC Summer 20236 upvotes
FlashCompact: A specialized model for context compaction
33k tok/sec context compaction for coding agents
Milestone
YC Summer 20235 upvotes
Reflexes API: Catch what agent logs miss
Most Agent failures are behavioral. Catch them at scale
Milestone
YC Summer 20232 upvotes
Kimi K3 on Morph: near-Fable performance at 1/5 the cost
Live on OpenRouter and Vercel's AI Gateway
Community
-
Community
10,500 tokens/SEC per request on Nvidia hardware
-
Community
Everything Is Model[S]
-
Community
Morph: Fast Apply AI Edits
-
Community
Show HN: Morph Reflexes – Multi-head classifiers for agent traces
-
Community
Optimizing Models to Be Fast at Codegen
-
Community
Morph Reflexes
-
Community
Show HN: Learning to test code changes with RL
-
Community
Show HN: Parallel Agentic Search on the Twitter Algorithm
-
Community
Show HN: Morph – Videos of AI testing your PR, embedded in GitHub
-
Community
WarpGrep – RL Subagent for Fast Context (Like SWE-Grep)