Recursive Superintelligence Releases Flint 40B MoE Coder Model with High SWE-bench Pro v2 Score

Recursive Superintelligence has launched Flint, a 39.8 billion-parameter Mixture-of-Experts (MoE) model designed for coding. The model, which features 2.5 billion active parameters and 256 routed experts, achieved an 89.7% score on SWE-bench Pro v2 and 74.2% on SWE-bench Verified, with Apache 2.0 weights and quantizations down to 11.4 GB.