HyLo: Long-Context Aware Upcycling for Hybrid LLMs Collection Hybrid MLA + Gated DeltaNet / Mamba-2 models from long-context aware upcycling: up to 32x usable context with over 90% KV-cache reduction. • 14 items • Updated 29 days ago • 2
Inkling MLX Collection MLX versions of Inkling 975B-A41B and 276B-A12B (Small) omni to text. • 6 items • Updated Jul 31 • 2
Wan-Streamer v0.1: End-to-end Real-time Interactive Foundation Models Paper • 2606.25041 • Published Jun 23 • 126
Ornith-1.0 Collection Ornith-1.0 is a family of open-source LLMs specialized for agentic coding. • 8 items • Updated 21 days ago • 396
Tiny Model, Big Logic: Diversity-Driven Optimization Elicits Large-Model Reasoning Ability in VibeThinker-1.5B Paper • 2511.06221 • Published Nov 9, 2025 • 142
view article Article The Open Source Community is backing OpenEnv for Agentic RL +18 burtenshaw, spisakjo, lysandre, darktex, willcb, qjoy, pawalt, cwing-nv, danielhanchen, andrewzhou, thegovind, shimmyshimmer, Hamid-Nazeri, Sanyam, zkwentz, emre0, lewtun, sergiopaniego, banghua, unseenmars • Jun 8 • 121
Qwen 3.x MTP Collection MLX MTP drafter checkpoints for Qwen 3.x speculative decoding with mlx-vlm. • 12 items • Updated Jun 1 • 9
SeedVR2 (MLX-Swift) Collection SeedVR2-3B (ByteDance, ICLR 2026) one-step diffusion super-resolution, MLX-Swift weights for on-device Apple Silicon. fp16 + int8. • 2 items • Updated Jun 6 • 3