Muse Spark OpenOpen Weight

Released

Muse Spark Open is an AI model released by Meta on Thursday, Oct 15 2026, 3 days after Muse Wonder Full. It is an open-weight model — the trained weights are available to download and run. It is a 816B parameter model with a 1M token context window. Benchmark results (shown below) cover MCP Atlas, DeepSWE 1.1, Terminal-Bench 2.1, JobBench, and GDPval-AA v2.

Benchmarks

Coding

DeepSWE 1.1Agentic coding — Artificial Analysis' independent test of deep, agentic software-engineering work — the AI has to plan and carry out substantial coding tasks end to end. (Version 1.1 of the test.) Higher is better.
60%
#22 of 35

Terminal & CLI

Terminal-Bench 2.1Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Higher is better.
76.4%
#22 of 35

Agentic & tool use

MCP AtlasMulti-step tool use — Can the AI chain together many tools and steps to complete one bigger task, rather than doing just a single thing? Higher is better.
89.2%
#1
JobBenchProfessional tool use — Tests the AI on professional workplace tasks that require using real work tools — the kind of multi-step jobs an office worker handles. Higher is better.
61.8%
#5 of 11

Knowledge work

GDPval-AA v2Knowledge work — economically valuable knowledge work (v2, re-based Elo)
1401
#17 of 21

Compare Muse Spark Open with

Muse Spark Open

About

Muse Spark Open, released October 15, 2026, brought downloadable weights to Meta's Muse Spark line after its closed releases earlier that year. Meta positioned the natively multimodal model for coding, long-running agent tasks and multi-step reasoning, with 816B total parameters, 42B active and a 1M-token context window. The weights were released under the Muse Model License with limited commercial terms. Meta also announced hosted inference partners and support for inference engines including vLLM, SGLang and MLX, and demonstrated local agent workflows on two Apple M5 Ultra Mac Studios.

Meta's launch scorecard reported results at xhigh reasoning: 61.8% on JobBench, 89.2% on MCP Atlas, a GDPval-AA v2 Elo of 1401, 76.4% on Terminal-Bench 2.1 using Terminus 2, and 60% on DeepSWE 1.1 using Muse Code. The evaluation report identified the benchmark providers and harnesses behind the results; the launch table also included HLE-Diamond and tool-assisted visual tests, which use different question sets or tool conditions from the standard HLE and unaided BabyVision rows tracked here.

Frequently asked questions

Muse Spark Open was released by Meta on Thursday, Oct 15 2026.

All Meta releases

18 tracked