{"slug": "flint-a-small-moe-model", "title": "Flint: A Small MoE Model", "summary": "RecursiveSI released Flint, a 39B-parameter mixture-of-experts model with 2.5B active parameters per token, built from a 2-dense-layer, 47-MoE-layer architecture and trained from scratch through tokenizer, pretrain, midtrain, SFT and RL stages. Flint scores 54.9% on Terminal-Bench 2.1, 89.7% on SWE-bench Pro v2 and 74.2% on SWE-bench Verified, with the team reporting the full training pipeline was built in 4 months. Training from scratch gives more flexibility for downstream fine-tuning.", "body_md": "# Flint: A Small MoE Model\n\n[Flint](https://huggingface.co/RecursiveSI/flint) is a new MoE model: 39B parameters, 2.5B active per token. Architecture is 2 dense layers followed by 47 MoE layers.\n\nBenchmarks: 54.9% on Terminal-Bench 2.1, 89.7% on SWE-bench Pro v2, 74.2% on SWE-bench Verified.\n\nTrained from scratch (tokenizer → pretrain → midtrain → SFT → RL), giving more flexibility for fine-tuning downstream. The team built the complete training pipeline in 4 months.", "url": "https://wpnews.pro/news/flint-a-small-moe-model", "canonical_source": "https://julin.ai/2026/10/22/flint-moe-model/", "published_at": "2026-10-21 11:00:00+00:00", "updated_at": "2026-10-22 09:42:55.927127+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "large-language-models", "ai-research"], "entities": ["RecursiveSI", "Flint", "Terminal-Bench 2.1", "SWE-bench Pro v2", "SWE-bench Verified", "Hugging Face"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/flint-a-small-moe-model", "markdown": "https://wpnews.pro/news/flint-a-small-moe-model.md", "text": "https://wpnews.pro/news/flint-a-small-moe-model.txt", "jsonld": "https://wpnews.pro/news/flint-a-small-moe-model.jsonld"}}