whatmodel.sh / Recursive Flint

Recursive Flint

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Recursive Flint is an open-weights language model from Recursive, released October 2026. Its published size is 39.8B parameters (2.5B active per token) and a 262k-token context window. Licence: apache-2.0.

Catalogue identity notes

BF16 checkpoint; card says Transformers implementation not yet included. SGLang serving kit requires B300/H200. Generic autogenerated vLLM commands not treated as verified routes.

Model information sources

Release date ↗ · Weights licence ↗

No comparable score yet. Its published results are below, but they don't overlap enough with other models to calculate a score.

$ how it compares

Scored LLMs released closest to Recursive Flint, out of 1192 with a score.

#modelscorewhereRAMreleased
3GPT-6.1 Terra OpenAI84±1api $2.00/M2026-10
4Gemini 4 Carbon Google83±1limited preview2026-10
7Muse Wonder Full Meta82±2Muse Code only2026-10
14Kimi K3.1 Pro Moonshot AI80±2api price unknown2026-10
18Muse Wonder Meta78±1api $8.00/M2026-10

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasksnot ranked on this task yet · 2 tests
testresultvs the others in its rundifficultycountswhy
SWE-bench Verified lab74.2% ✎▁▁▁▁▁▁▂▅█▆▁▁ beats 98/136—not counted yet: no other model in the same run to compare it with
Terminal-Bench 2.1 lab54.9% ✎▁·▁·▁▁▄▃▆█▅█ beats 22/63—not counted yet: no other model in the same run to compare it with
codingnot ranked on this task yet · 1 test
testresultvs the others in its rundifficultycountswhy
SWE-bench Verified lab74.2% ✎▁▁▁▁▁▁▂▅█▆▁▁ beats 98/136—not counted yet: no other model in the same run to compare it with

$ where to run it

No verified API or download recorded yet · add one

Official weights and documentation: huggingface.co/RecursiveSI ↗

$ published results · 2 results, not enough overlap for a score
testresultsource and conditions
Terminal-Bench 2.154.9%Recursive Flint Apache-2.0 model card · lab-reported

Recursive own BF16 release figures; temperature 1, top-p .95, top-k 20; average pass@1 over at least five attempts. Terminus 2 for Terminal-Bench, excluding solutions from web retrieval; mini-swe-agent for SWE. SWE-Pro v2 excluded because no matching definition exists.

SWE-bench Verified74.2%Recursive Flint Apache-2.0 model card · lab-reported

Recursive own BF16 release figures; temperature 1, top-p .95, top-k 20; average pass@1 over at least five attempts. Terminus 2 for Terminal-Bench, excluding solutions from web retrieval; mini-swe-agent for SWE. SWE-Pro v2 excluded because no matching definition exists.

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json