AI
Recursive Unveils Flint, a High-Speed Local Coding Model
Recursive has launched Flint, an open-weight 40B MoE model for coding agents that achieves up to 150 tokens per second on local hardware.
October 22, 2026, 01:39 AM·MarkTechPost

Recursive has launched Flint, an open-weight 40B MoE model for coding agents that achieves up to 150 tokens per second on local hardware.
