Recursive's Automated Research System Produces Flint, a Self-Improving Sparse Model

Recursive SI aims to automate the full research loop, and its Eureka Machine system has trained Flint, a 40B-parameter sparse model with only 2.5B active parameters that runs agent coding locally at 125+ tokens/s on an M4 Pro.

2026-11-12 ~ 2026-11-12 · 4 related posts

• Recursive debuts Flint, the first model from its auto-research system, at Gemma hardware hackathon — agihouse_org · 2026-11-12
• Recursive plans to automate the full research loop, from GPU clusters to wet labs — agihouse_org · 2026-11-12
• Recursive's Eureka Machine ships Flint: a sparse 40B model with 2.5B active for local agentic coding — agihouse_org · 2026-11-12
• Open-source sparse model Flint: 40B params, 2.5B active, 125+ tok/s on M4 Pro for local agentic coding — agihouse_org · 2026-11-12