Recursive AI launches Flint: a 40B MoE local coding model hitting 125+ tokens/s on MacBook Pro

_rockt · x · 2026-10-22

Recursive AI, founded by Richard Socher, released Flint, its first model built from scratch in under 4 months by the company's automated research system, the Eureka Machine.

• Specs: 40B MoE with 2.5B active params, 262K context window, Apache 2.0 open weights
• Positioning: a fast local agentic coding model — 125+ tokens/s on a MacBook Pro and 150+ on RTX Spark, up to 2x faster than comparable models
• Deployment: runs locally on 24GB, day-0 support in LM Studio, free to try on opencode and pidot.dev

The pitch: private, unmetered intelligence on your own machine, with impressive benchmark performance.

Related event: Recursive releases Flint, an open-weights 40B MoE coding model built in under 4 months by its automated Eureka research system(49 posts)→

Original post →

More from coding & agent

• How a second Mac mini lets AI agents auto-distribute compute-heavy tasks — tedddyoweh · 2026-10-22
• LangChain's Design Agent Shipped 3,000+ Assets and Earned 4.5M Views Since August — LangChain · 2026-10-22
• Databricks Chief AI Scientist: Evals Are the Biggest Bottleneck to Enterprise AI Adoption — jefrankle · 2026-10-22
• Developer plans a solar-aware job scheduler so his homelab rides free excess power — mertdumenci · 2026-10-22
• How do you design your agentic code review workflow? — lucasmeijer · 2026-10-22
• Langfuse now supports custom System One decision models including SubQ Snap — ryanallred · 2026-10-22

coding & agent channel →