Meta's 816B open-weight Muse Spark Open hits Groq preview at 350+ tokens/sec
GroqLLC · x · 2026-10-16
Groq announced limited-preview hosting of Meta's open-weight Muse Spark Open at 350+ tokens/sec per user.
- Specs: 816B-parameter sparse MoE activating 42B per token; text/image/video/audio input; 1M-token native context (131,072 on GroqCloud); trained for long-horizon agentic work like coding and customer-service agents. Released under the Muse Model License.
- Ecosystem: runs on SGLang, vLLM, TensorRT-LLM, Transformers, MLX, llama.cpp, Unsloth; verified on NVIDIA, AMD, Intel and Apple silicon; BF16 and NVFP4 precisions plus FP8/MXFP4/INT4 quants; hosted by Baseten, Fireworks, Together AI and more; fine-tuning via TorchTitan, NeMo, TRL live at launch.
- Meta is also running a Global AI Developer Hackathon with $1M in prizes.
More from Infra
- Berkeley scholars: concentrated chip supply chain makes a hardwired frontier AI pause feasible — wfithian · 2026-10-16
- Microsoft trained MAI-Base-2 on 16k GB200 GPUs in just 38 days, with 21% MFU — scaling01 · 2026-10-16
- Robotics data tokenization landscape: Khala maps peaq, THOR, BitRobot and Axis — 0xSammy · 2026-10-16
- Signal65 benchmarks: Intel Xeon 6 runs agent workloads 11% faster for less money — ryanshrout · 2026-10-16
- Microsoft's MAI-Base-2: 2.8T-param MoE pretrained on 16k GB200s in just 38 days — scaling01 · 2026-10-16
- Cloudflare and OpenAI pilot: real-time network signals bring unseen pages into ChatGPT search — irvinebroque · 2026-10-16