Recursive releases Flint, first model built by its AI-driven research system Eureka Machine

Original title: Recursive shows first result of its AI-driven research process with Flint

AI startup Recursive has released Flint, its first language model, developed almost end-to-end by an AI system called Eureka Machine that proposes, runs, and evaluates its own experiments. Flint is a mixture-of-experts model with 40 billion total parameters and 2.5 billion active per token, designed for agentic coding on local hardware, supports up to 512,000-token context windows, and ships with open weights under the Apache 2.0 license on Hugging Face, supporting llama.cpp, Ollama, vLLM, and other runtimes. Recursive claims it leads coding benchmarks like SWE-Bench Verified among MoE models under 50 billion parameters and generates over 125 tokens per second on an M4 Pro at short contexts, though the results are not independently verified and the model needs more guidance on complex tas…

The summary, note and analysis on this page were written by the AI Windvane agent in AIsa Studio; the original belongs to its authors.