Meta releases Muse Spark Open, an 816-billion-parameter open-weight model that still trails China's Kimi K3 on coding

Meta Superintelligence Labs put the weights of Muse Spark Open on Hugging Face on October 15. It is the first open release from the Muse family. Until now, Meta had kept Muse closed and offered it only through its own API and apps. The model card lists 816 billion total parameters, with 42 billion active per token. Reporting on the release describes a mixture-of-experts Transformer with 256 experts, 8 of them routed per token. The model has a 1-million-token context window, accepts text, image, video and audio, and outputs text. It shares a base model with the closed Muse Spark 1.3 and received extra safety post-training so it can run without Meta's system-level guardrails.

Meta ships two weight formats. One is BF16. The other is NVFP4, NVIDIA's 4-bit floating-point format, which Blackwell hardware supports natively. Serving NVFP4 takes at least about 700 GB of accelerator memory, and BF16 takes about 2 TB. The most notable deployment claim is a desktop setup: two M5 Ultra Mac Studios, each with 512 GB of unified memory. VentureBeat notes that this memory configuration has not shipped publicly yet. CNBC described the work as a joint Apple and Meta effort to run the model on Macs. A Russian-language AI channel called it one of the strongest open models in the United States and noted that it runs without cloud servers.

The benchmark results are less favorable than the launch presentation suggests. A comparison of Meta's published numbers puts Muse Spark Open at 76.4 on Terminal-Bench 2.1, against 85.0 for Moonshot AI's Kimi K3. On SWE-Atlas-QnA it scores 58 to Kimi K3's 66. It also trails Qwen, GLM and Kimi configurations on Terminal-Bench 2.1, and on DeepSWE 1.1 it beats Qwen but trails GLM and Kimi. It does beat the other US open models, NVIDIA's Nemotron 3 Ultra and Thinking Machines' Inkling, often by more than 20 points. Meta's model card claims a score of 39 on the Artificial Analysis Intelligence Index. VentureBeat could not verify any independent Artificial Analysis assessment at publication. If the score is confirmed, it would tie DeepSeek V4.1 Flash for fifth place.

The license is the main practical constraint. The Muse Model License closely resembles Kimi K3's license, with commercial limits and naming requirements. Alibaba's Qwen and Zhipu's GLM have mostly shipped under Apache 2.0 or MIT, which have no revenue thresholds.