Recursive's Eureka Machine ships Flint: a sparse 40B model with 2.5B active for local agentic coding
agihouse_org · x · 2026-11-12
At the Google Gemma Hardware Hackathon, Recursive SI co-founder Tim Shi unveiled Flint, the first model built by their auto-research system (Eureka Machine):
- Architecture: sparse 40B-parameter model with only 2.5B active, built for local agentic coding
- Performance: 125+ tokens/sec on an M4 Pro MacBook Pro
- Key tech: a custom Metal megakernel for Apple M4/M5 chips, generated by the auto-research system
- Open weights
Shi also outlined the layers of recursive self-improvement: learning from data → RL → learning the reward itself → discovering new architectures. Every layer can be written as code, and AI has become very good at writing code.
More from Infra
- Sixth Street CEO: ~$200B of Chips Sitting in Warehouses as Data Centers Lag Behind — SumitGup · 2026-11-12
- AWS CEO Matt Garman: Space Data Centers Are 10-20 Years Away, Likely Not Economically Feasible — rohanpaul_ai · 2026-11-12
- CoreWeave Database Goes GA: Managed PostgreSQL Running Inside Your CKS Cluster Next to GPUs — _ScottCondron · 2026-11-12
- micro1 scaled eval infra to 12× monthly tokens in 5 months, agent rollouts now start in 12 seconds — Exp_Mark · 2026-11-12
- Big AI conflates scale with innovation as voters reject data center ballot measures — DavidLinthicum · 2026-11-12
- Musk says SpaceX targets 1TW/year from Earth, rising to 1PW/year from the Moon — ns123abc · 2026-11-12