Recursive Superintelligence has launched Flint, a 40B parameter Mixture-of-Experts (MoE) model designed for coding agents. This model, which uses 2.5B active parameters per token, was developed with the assistance of Recursive's automated research system, Eureka Machine. Flint demonstrates strong performance on coding benchmarks like IFEval and SWE-bench, and is optimized to run efficiently on Apple Silicon Macs and NVIDIA GPUs with 48GB of memory. AI
IMPACT Optimized for local execution on consumer hardware, this model could accelerate agent development and deployment for developers with limited resources.
RANK_REASON Frontier-lab model release with system card [lever_c_demoted from frontier_release: ic=1 ai=1.0]
- Apache Software License 2.0
- Apple M4 Pro
- Apple M5 Max
- Apple M5 Pro
- Apple Silicon
- Eureka Machines
- Hugging Face
- IFEval
- Nvidia
- Qwen
- Recursive Superintelligence
- SWE-bench
- Terminal-Bench
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →