Recursive Debuts Open-Weight Flint Model for Fast Local Coding Agents
AI startup Recursive has released Flint, a 40-billion-parameter open model optimized to execute complex agentic coding tasks locally on a laptop.
Artificial intelligence startup Recursive launched Flint, an open-weight mixture-of-experts model specifically architected for localized software development workflows. The software activates just 2.5 billion parameters per token, enabling individual laptops to process repository tasks, bug investigations, and code generation entirely offline.
The engineering milestone highlights the potential of automated development loops, as the model was largely generated by Recursive's own autonomous AI pipeline. By bypassing third-party API costs, independent developers gain access to fast private environments that heavily outperform comparable sub-50-billion parameter architectures.
Original source
techgig.com
Related Articles
Jack Dorsey's Block Introduces Local Speech-to-Speech Voice Model Dessa
Jack Dorsey's financial technology company Block has revealed Dessa, a text-free voice AI model that runs locally on a laptop.
Microsoft Diagnostic AI Diagnostic Tool Defeats Human Doctors
A new Microsoft sequential artificial intelligence system has surpassed human medical professionals in diagnostic accuracy during a benchmark study.
Microsoft AI Model Outperforms Human Doctors in Medical Diagnosis Benchmark
A study published in Science reveals that Microsoft's sequential AI diagnostic tool achieved 83.2% accuracy, beating general physicians.
Block Unveils Dessa, a Voice AI That Talks and Listens at the Same Time
Jack Dorsey's Block built Dessa, a voice model that can speak and listen simultaneously and runs locally on consumer hardware.