An independent research notebook on AI engineering, agents, models and the systems around them.https://julin.ai/atom/tags/models/index.xmlJu Lin2026-10-22T00:00:00+13:00https://julin.ai/2026/10/22/flint-moe-model/2026-10-22T00:00:00+13:002026-10-22T00:00:00+13:00<p><a href="https://huggingface.co/RecursiveSI/flint">Flint</a> is a new MoE model: 39B parameters, 2.5B active per token. Architecture is 2 dense layers followed by 47 MoE layers.</p> <p>Benchmarks: 54.9% on Terminal-Bench 2.1, 89.7% on SWE-bench Pro v2, 74.2% on SWE-bench Verified.</p> <p>Trained from scratch (tokenizer → pretrain → midtrain → SFT → RL), giving more flexibility for fine-tuning downstream. The team built the complete training pipeline in 4 months.</p>