π Qwopus3.6 Collection This collection features the advanced Qwopus3.6 series of multimodal large models, which are fine-tuned from the Qwen3.6 base models with a focus on e β’ 4 items β’ Updated 3 days ago β’ 36
Mistral Medium 3.5 Collection Our first flaship models handling instruction-following, reasoning, and coding in a single set of opened-weights. β’ 2 items β’ Updated 11 days ago β’ 16
MT5 release Collection The MT5 release follows the T5 family, but is pretrained on multilingual data. The update UMT5 models are pretrained on an updated corpus. β’ 10 items β’ Updated Mar 12 β’ 25
Moonlight-A3B Collection Moonshot's Compute-efficient MoE LLM, first Scaling Up of Muon Optimizer β’ 3 items β’ Updated Jan 27 β’ 12
Whisper Release Collection Whisper includes both English-only and multilingual checkpoints for ASR and ST, ranging from 38M params for the tiny models to 1.5B params for large. β’ 12 items β’ Updated Sep 13, 2023 β’ 157
gpt-oss-safeguard Collection gpt-oss-safeguard-120b and gpt-oss-safeguard-20b are safety reasoning models built-upon gpt-oss β’ 2 items β’ Updated Oct 29, 2025 β’ 72
gpt-oss Collection Open-weight models designed for powerful reasoning, agentic tasks, and versatile developer use cases. β’ 2 items β’ Updated Aug 7, 2025 β’ 434
Distilled Models Collection Kimi-2.5, Gemini 3.1 Pro, Qwen3.6-plus β’ 9 items β’ Updated 22 days ago β’ 7
Alchemist Collection π Dataset and π checkpoints for paper π "Alchemist: Turning Public Text-to-Image Data into Generative Gold" β’ 8 items β’ Updated Oct 16, 2025 β’ 18
Trinity Collection Collection of Arcee AI models in the Trinity family β’ 14 items β’ Updated Mar 25 β’ 30