MatBrain splits reasoning from tool use: two models screen 30,000 crystal candidates in 48 hours
bravo_abad · x · 2026-09-23
Shi and coauthors argue that forcing one large model to master both scientific reasoning and tool orchestration may be the wrong architecture. Their MatBrain system splits the job between two specialized models:
- Mat-R1: a 30B-parameter model for materials reasoning
- Mat-T1: a 14B model trained to plan calculations, call tools, handle errors, and decide next steps
The two models repeatedly exchange results and instructions. The rationale: scientific analysis has relatively constrained answers, while tool use requires exploring many action sequences, arguments, and recovery paths. The system screened 30,000 crystal candidates in 48 hours, demonstrating a dual-model agent architecture for scientific discovery.
More from coding & agent
- LangSmith Adds Support for Decision Models Jev and SemIf for Step-Level Debugging — LangChain · 2026-09-23
- mcp-server-github-gist: MCP server to manage GitHub Gists from your IDE — modelcontextprotocol · 2026-09-23
- Opus 5.5 Hand-Draws a Starling Murmuration; Owner Finds Models Never Sign Their Handoff Notes — RileyRalmuto · 2026-09-23
- Stripe ships WebMCP for browser agents: 42% fewer tokens, 38% fewer tool calls — jeff_weinstein · 2026-09-23
- Swarms Marketplace adds private GitHub repo import for listing agents for sale — KyeGomezB · 2026-09-23
- Making Opus 5 write its own handoff prompt before Opus 5.5 kills its workers — doodlestein · 2026-09-23