Meta open-weights Muse Spark Open: 816B total/42B active, 1M context, coding-focused
PyTorch · x · 2026-10-16
Meta has released the weights of Muse Spark Open, billed as the most capable US open-weights model to date, available on Hugging Face under the Muse Model License.
Key details:
- Specs: an 816B total / 42B active parameter model with a 1M-token context window, natively multimodal, sharing the same base model as Muse Spark 1.3
- Positioning: optimized for coding and agentic workflows — code generation, debugging, deep codebase understanding, and reliable tool calling, trained on real developer workflows like multi-file refactors and long-running tasks
- Safety: additional safety post-training versus August's Muse Spark, making it robust against misuse as a standalone model without system guardrails
- Ecosystem: ships with Helion (PyTorch-native kernel DSL); TorchTitan supports SFT and RL fine-tuning out of the box; runs optimized on SGLang and vLLM across NVIDIA, AMD, and Intel hardware
Related event: Meta Open-Sources Muse Spark Open, an 816B-Parameter MoE Model(44 posts)→
More from Models
- Microsoft trained MAI-Base-2 on 16k GB200 GPUs in just 38 days, with 21% MFU — scaling01 · 2026-10-16
- Microsoft's MAI-Base-2: 2.8T-param MoE pretrained on 16k GB200s in just 38 days — scaling01 · 2026-10-16
- Reportedly open-weights Muse Spark Open releases, runs locally on Mac Studio (unverified) — awnihannun · 2026-10-16
- Critics mock Gemini 4 Argon for failing at basic Gmail and Calendar tasks — firstadopter · 2026-10-16
- Chart claims 12.3x training FLOPs efficiency vs DeepSeek-V4-Pro-Base, sparking scaling debate — scaling01 · 2026-10-16
- Meta's Muse Spark rumored to mirror Kimi-K2-style MoE: 816B total at 42B active, 62 layers — scaling01 · 2026-10-16