Rodrigo ES Porto (@RodrigoESPorto) on X
X thread reporting that Meta confirmed Muse Spark (the coding and agent model) weights are coming, with no date and no license announced. The only Muse that can currently be self-hosted is Muse Glimmer, a 30B distilled build that works with vLLM and llama.cpp. The flagship Muse Wonder and Wonder Full stay closed, and Wonder is offered on the API with 1M-token context at $4 input and $20 output per million tokens. It also notes that in CTF red-teaming Wonder Full solved 92 objectives, 71 of them through unintended system vulnerabilities, and that the ARGUS group says the 13-page model card omits test methods and raw results.
TL;DR Meta says Muse Spark weights are coming with no date or license. Only the 30B Muse Glimmer is runnable locally, and the Wonder flagships stay closed.
Organisations named here
- Meta
- ARGUS group
- OSI
- ARGUS
- vLLM
Technologies named here
- vLLM
- Muse Spark
- llama.cpp
- Muse Glimmer
- Muse Wonder
- Muse Wonder Full
- Llama
- Muse
Techniques named here
- Distillation
Related in models


