Socher's Recursive Unveils Flint: 125+ tok/s Local Coding Model Built by Automated Research
jeffclune · x · 2026-10-28
Recursive has released its first model, Flint: a fast local coding model built by its automated research system, the Eureka Machine.
- Best-in-class fast local coding, up to 2x faster than comparable models — 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark.
- Open weights available on Hugging Face; runs locally on your favorite framework or free via opencode or pi.
- Built from scratch in under 4 months.
In the same talk, Socher explained why open-endedness is crucial for AI breakthroughs: "if you only focused on improving fire, you'd never invent the microwave" — going directly after the goal makes you miss clever ideas. He cites the Darwin Gödel Machine paper (four of five authors now at Recursive) as originating the plots later popularized by Karpathy's Auto Research, and notes open-ended approaches apply even to prompt injection research, e.g. having one LLM attack another.
Related event: Recursive Unveils Flint, an Auto-Research-Built Local Coding Model(2 posts)→
More from Models
- OpenBMB releases open-source MiniCPM-OCR 1.0 with online demo — Dr_Singularity · 2026-10-28
- Liquid AI LFM3's hybrid architecture: 6 global attention layers, 256K context, 25% faster inference — cneuralnetwork · 2026-10-28
- Dev names GLM 5.5 Flash and Qwen 4 27B as the models he's most excited about — victormustar · 2026-10-28
- Maxime Labonne demos d1: bounding boxes for named objects in a single forward pass — maximelabonne · 2026-10-28
- mlx-vlm Will Stay Open Source Under Same License, Only Moving to Org Repo — andrejusb · 2026-10-28
- Users report Grok Bot latency dropped sharply, with noticeably snappier responses — mark_k · 2026-10-28