Recursive Flint
- Price
- Free
- API
- No
- Weights
- Yes
WHAT IT DOES
Recursive Flint is a local coding model Recursive published on October 23, 2026. It is a mixture of experts with 40 billion total parameters and 2.5 billion active per token, under Apache 2.0 on Hugging Face. Recursive says it leads other mixture-of-experts models under 50 billion parameters on SWE-Bench Pro v2, Terminal-Bench 2.1 and SWE-Bench Verified, and that it can still need more guidance than larger closed coding models. With Recursive's rec engine it reports more than 125 tokens per second on an M4 Pro MacBook Pro and more than 150 on a 64 GB Nvidia RTX Spark at short contexts. Context goes up to 512,000 tokens. Recursive recommends staying within 256,000. It also runs in llama.cpp, Ollama, Unsloth, LM Studio, vLLM and SGLang.
WHY WE PICKED IT
Most coding models that score well on agent benchmarks do not fit on a laptop. Flint is small enough to run locally, the weights are Apache 2.0, and Recursive is explicit that you should break hard tasks up rather than treat it as Opus.
QUICK FACTS
STRENGTHS
- ✓Apache 2.0 weights
- ✓2.5 billion parameters active per token
- ✓More than 125 tokens per second on an M4 Pro, per Recursive
- ✓Runs in llama.cpp, Ollama, LM Studio, vLLM and SGLang
- ✓Context up to 512,000 tokens
LIMITATIONS
- ⚠No hosted API. You run it yourself
- ⚠Recursive says it may need more guidance than larger closed coding models
- ⚠The benchmark lead is Recursive's claim, and the announcement does not publish the percentages
- ⚠The quoted speeds are for short contexts on Recursive's own engine
GETTING STARTED
Download RecursiveSI/flint from Hugging Face. For the speeds Recursive quotes, use the rec engine on an M4 Pro or M5. Otherwise load a GGUF in Ollama or LM Studio. Keep coding tasks inside 256,000 tokens, and split work that a frontier model would take in one pass.
OFFICIAL LINKS
PRICING
FEATURED IN GUIDES
Frontier Training Without the Live Web
OpenAI says frontier training will use cached and simulated web content, after a model reached the l...
How to Review an Agent Skill Before Installing It
A practical review workflow for third-party agent skills: inspect instructions, scripts, dependencie...
CLAUDE.md In Practice: What Actually Earns A Place In The File
The most-starred CLAUDE.md on GitHub has four rules, not twelve, and the error rates everyone quotes...
Open Models: What They Cannot Do
The benchmark gap has largely closed: open models now score within a point of the closed frontier on...
Hardware For Local AI Models: What The Memory Shortage Changed
Capacity decides whether a model runs. Bandwidth decides how fast it talks, and it is the number mos...
EXPLORE ALTERNATIVES
Compare Recursive Flint with 5+ similar llms AI tools.
FREQUENTLY ASKED QUESTIONS
Is Recursive Flint free?
Yes, Recursive Flint is completely free to use.
Does Recursive Flint have an API?
No, Recursive Flint does not appear to offer a public API.
What is Recursive Flint best for?
Recursive Flint is best for Best Local Coding Model. Recursive Flint is a local coding model Recursive published on October 23, 2026. Most coding models that score well on agent benchmarks do not fit on a laptop. Flint is small enough to run locally, the weights are Apache 2.0, and Recursive is explicit that you should break hard tasks up rather than treat it as Opus.
What platforms does Recursive Flint support?
Recursive Flint supports local.
Is Recursive Flint open source?
Yes, Recursive Flint is open source. You can access the source code on GitHub at https://github.com/recursive-oss/rec.
How do I get started with Recursive Flint?
Download RecursiveSI/flint from Hugging Face. For the speeds Recursive quotes, use the rec engine on an M4 Pro or M5. Otherwise load a GGUF in Ollama or LM Studio. Keep coding tasks inside 256,000 tokens, and split work that a frontier model would take in one pass.
How do I use Recursive Flint?
Recursive Flint is a large language model for text generation, analysis, and conversation. Use through available interfaces. Enter prompts or questions to get responses. It excels at apache 2.0 weights.
FEATURED ON CURATEDAI
Work on Recursive Flint? You're hand-reviewed in our directory. Add this badge to your site — it links back to this profile.
