AI startup Recursive launched Flint on Wednesday, October 21, introducing a free coding model designed to run directly on laptops. The company says its automated research system, Eureka Machine, helped build nearly every part of Flint in under four months. The release puts Recursive’s vision of using AI to accelerate the development of more capable AI systems to the test.
Flint targets developers who want coding assistance without relying on cloud services or transmitting source code to external servers. The company released the model under the Apache 2.0 license, allowing developers to download, modify and deploy it. Recursive says the model can work offline and avoids usage fees tied to cloud-based coding tools.
Flint Combines Large Scale with Local Performance
Flint uses a mixture-of-experts architecture with 40 billion parameters, activating only 2.5 billion for each token. This design reduces the computing demand and allows the model to run on compatible consumer hardware. Recursive reports speeds of around 131 tokens per second on a MacBook Pro with Apple’s M4 Pro chip.
The company says Flint outperformed other models with fewer than 50 billion parameters on SWE-Bench Verified, a software bug-fixing benchmark. Its internal evaluation across six tests also placed Flint behind only Alibaba’s Qwen3.8 27B. However, Recursive acknowledges that developers may need to divide complex coding jobs into smaller tasks.
Eureka Machine Drives AI Research
Recursive says Eureka Machine generated ideas, tested approaches and helped optimize Flint’s architecture, training and inference. Its contributions reportedly include low-bit training that reduced the model’s size by 71% with minimal quality loss. Human researchers still supervised the process and selected promising results.
Recursive emerged from stealth in May with USD 650 million in funding and a reported USD 4.65 billion valuation. CEO Richard Socher, Salesforce’s former chief scientist, leads the company’s push toward AI-assisted research.
Flint offers developers a practical local coding tool while advancing Recursive’s larger experiment. The company ultimately wants successive AI models to improve its research system and accelerate future model development.
Also Read: India’s AI Strategy 2027: Opportunities in Research, Infrastructure, Startups and Governance