Re
Recursive: Flint
Flint is an open-weight mixture-of-experts model from Recursive, with about 2.5B active parameters out of 40B total, built by Recursive's autoresearch system Eureka Machine for coding agents. It is suited...
- Input
- text
- Output
- text
Providers
Available through 3 free providers. Use the model ID and API settings below to connect.
* The provider match may differ from this model. Check its docs for details.
Benchmarks
Compare scores and explore results by category.
No benchmark results are available for this model yet.
Frequently asked questions
Free access, setup, and what to expect.
Flint is available with up to 200 requests / hour free through Kilo. Free allowances can change. Check the Flint provider limits before you start.
To use Flint, choose one of the listed Flint API providers and follow the provider’s account and API-key setup guide. Copy the Flint model ID and connection settings for your chosen provider into your app or API client. Model IDs can differ between providers. API requests go directly to the provider; Free Models is a comparison site.
Flint accepts text as listed input formats. Listed output formats include text. Image or audio input support does not imply image or audio generation. Check the chosen provider’s Flint documentation for supported file types and request formats.
Flint has a listed context window of 262K tokens. Tokens are small pieces of content rather than a word count. Instructions, conversation history, and other inputs use context space, and the response also needs room. The maximum output length for Flint may be smaller than the context window and can vary by provider.
Flint has both reasoning and tool calling listed as supported capabilities. Reasoning enables additional processing for complex problems. Tool calling allows a model to request functions that your app executes. Check the chosen provider’s documentation for capability availability and setup.
No benchmark results are listed for Flint yet. Missing scores don’t establish poor performance. Compare scores within the same benchmark: coding indexes, Elo ratings, and win rates measure different things. The index comparison charts include only scored free models listed on Free Models. Evaluate Flint on examples from your own task alongside context size and provider limits.