Recursive releases Flint, a coding model developed with its Eureka Machine AI system

What happened

Recursive released Flint, an open-weight MoE language model for local agentic coding, developed in four months with its Eureka Machine AI research system. Flint has 40 billion total parameters and 2.5 billion active per token, with Apache 2.0 weights available on Hugging Face. Recursive says the system contributed across architecture, training, quantization, and inference tuning; its benchmark claims have not been independently verified, and Flint does not yet demonstrate recursive self-improvement.

Read next

• More changes in today’s briefingMicrosoft Research and Collaborator Resolve the KLS Conjecture With Human-Directed AI

Reported by 2 sources

The story so far

Recursive released Flint, a coding model for local use on laptops, with weights available free on Hugging Face under the Apache 2.0 license, which permits use and modification. Recursive says a small team built it from scratch in about four months with help from its Eureka Machine AI research system, which contributed to architecture, training, quantization and inference tuning. Flint has 40 billion total parameters and activates 2.5 billion per token. Recursive reports generation speeds of about 131 tokens per second on an M4 Pro MacBook Pro, while acknowledging that Flint needs more guidance than large paid coding assistants. The Decoder reports that Recursive’s benchmark claims have not been independently verified and that Flint does not yet demonstrate recursive self-improvement.