Oriel — AI & technology worth noticing, every day.

unsloth/flint-GGUF

AI model · https://huggingface.co/unsloth/flint-GGUF

First seen by Oriel on .

AI summary

Flint is Recursive’s 39.8B-parameter mixture-of-experts model for coding agents, with release evaluations reporting strong results on Terminal-Bench and SWE-bench.

  • The decoder-only model activates approximately 2.5B parameters per token and supports reasoning, multi-turn tool use, and text input and output.
  • It scored 54.9% on Terminal-Bench 2.1, 89.7% on SWE-bench Pro v2, and 74.2% on SWE-bench Verified; evaluation setups vary across sources.
  • The repository includes a BF16 checkpoint, model configuration, tokenizer, chat template, and SGLang serving kit; quantized variants target NVIDIA and Apple Silicon runtimes.
  • The documented context configuration is 262,144 tokens, shared between prompts and generated output; long reasoning can increase latency or exhaust the output budget.
  • Released under Apache 2.0, Flint may produce faulty code or unsafe tool calls, so applications should test outputs and enforce their own permissions.

#mixture-of-experts #coding agents #llm inference #gguf

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Likes

Total likes reported by the source.

  • Hugging Face Models: 345, latest observation ; +233 over observed period since

Downloads (30d)

Rolling 30-day downloads. A change in this value is not new downloads during the observed period.

  • Hugging Face Models: 232,764, latest observation ; +232,753 over observed period since

Aliases

SourceExternal IDURL
Hugging Face Modelsunsloth/flint-GGUFhttps://huggingface.co/unsloth/flint-GGUF Read discussion