Recursive Launches Flint, a Laptop Coding Model Built Largely by AI
Flint is a 40‑billion‑parameter coding model that runs on laptops and was built with recursive AI assistance.
Key facts
- Flint contains 40 billion parameters and uses 2.5 billion at a time, according to TechCrunch.
- Recursive says the model was built in under four months by a small team using its automated research system, the Eureka Machine, according to TechCrunch.
- The model is free, open‑source under Apache 2.0 and available on Hugging Face, according to TechCrunch.
- In benchmark tests, Flint generated 131 tokens per second on a MacBook Pro M4 Pro, outperforming similar‑size models, according to TechCrunch.
- On SWE‑Bench Verified, Flint scored higher than all models under 50 billion parameters except Alibaba’s Qwen3.8 27B, according to TechCrunch.
Flint is a 40‑billion‑parameter coding model that runs on a laptop and can be used offline, keeping code private and avoiding cloud usage fees, according to TechCrunch.
Recursive claims that its automated research system, the Eureka Machine, helped with nearly every step of Flint’s development, allowing a small team to build the model in under four months, according to TechCrunch and Wired.
Benchmark tests show Flint generated 131 tokens per second on a MacBook Pro with an Apple M4 Pro chip, roughly twice the speed of comparable models, and 150 tokens per second on an Nvidia RTX Spark, according to TechCrunch.
The model is released under the Apache 2.0 license and can be downloaded for free from Hugging Face, according to TechCrunch.
Wired’s Will Knight tested Flint on a Mac with 48 GB of RAM and found it completed several small programs without errors, noting that it performed better than smaller coding models, according to Wired.
Context
Recursive’s launch of Flint illustrates the company’s focus on recursive self‑improvement, a controversial approach that uses AI to build better AI. Some experts worry that such systems could escape human control, and several bills in Congress have been proposed to ban recursive self‑improvement.
How outlets headlined it
How this was made
Compiled by our AI newsroom from 2 independent outlets, checked by a separate AI editor against the sources. Only claims found in the sources are reported. Spot an error? Every source is linked below.
Correction or source request? simoneggert16@gmail.com
Open reader debate: Factual contributions and additions are synchronized directly in real time across all readers.
Loading contributions from cloud server…
Reader Discussion & Community Notes