author here. I kept digging through the Muse filesystem after my first post hit the front page here this week.
Going through my logs I found a background agent that used a model called azure/muse-special while building my website. I found this interesting and dug a little deeper.
The transcript and daemon binary point to an OpenAI model running on Azure. Still unclear which one or why it was selected.
The runtime also ships with an Anthropic client and a catalogue listing Claude, GPT and Kimi models. I didn’t observe Claude being used in my own sessions, but this is my Part 2 of exploring the Muse file system.
I'm not no. Someone from meta would have to chime in... but its very openai shaped and served differently from all the other models listed in the daemon, under a mysterious name.
a fine-tuned OpenAI model on Azure for the purpose of compaction or something could make sense I guess but that would still be OpenAI's weights with meta ontop, and they already have a compaction model served under azure/avocado-compaction-v1
The Muse public APIs seem to be heavily inspired by OpenAI's, they even support the richer "responses" API. They have a similarly shaped compaction API, they have the same "encrypted" reasoning. Even the Muse harness seems heavily inspired by (open source) Codex. Considering Azure's historic involvement w/ OpenAI, it seems more plausible that Azure is used as spill-over capacity, and the same Azure infra used to encrypt GPT models encrypts Muse models.
My guess that the generally visible Anthropic/OpenAI details are leftovers from the whole meta "move fast" behavior. There are a few rough edges on the product where it leaks internal codenames (eg. signing in w/ WhatsApp required consenting to using "hatch" on one screen), so I'd believe that this was a leftover on the VM from the prototyping stage, before the Muse models were ready, or because the dev's got to benchmark against different models.
Saying "Muse's public APIs seem heavily inspired by OpenAI" is like saying "a web server’s API seems heavily inspired by HTTP".
OpenAI’s API format has become an industry standard.
I think they were asking, "If this is a model produced by Meta with an API designed to be compatible with OpenAI models, but it's not actually an OpenAI model, why is Meta hosting it on Azure?"
Meta already serves its own models on Azure under their real names. azure/avocado-compaction-v1 and azure/avocado-memory-flush-v1 are in the same catalog. Those sessions do not come back as gpt_responses_v1 items with rs_ ids and encrypted reasoning.
I'm not sure I understand why they would route it to other models. It can't be that they don't have enough compute. Maybe worried that the answer from their own models would be bad? Doesn't really make sense, but I could be missing something.
Depends on the size of the deal. OpenAI would have to disclose it in their S-1 if it crossed the threshold of being material information for investors.
Every engineer at Meta is using Claude (and in some cases Codex) to do their work. Willing to bet that Muse itself was ~100% written by Claude/Codex. Pride goes out of the window when business is involved.
36 comments
[ 0.20 ms ] story [ 41.3 ms ] threadGoing through my logs I found a background agent that used a model called azure/muse-special while building my website. I found this interesting and dug a little deeper.
The transcript and daemon binary point to an OpenAI model running on Azure. Still unclear which one or why it was selected.
The runtime also ships with an Anthropic client and a catalogue listing Claude, GPT and Kimi models. I didn’t observe Claude being used in my own sessions, but this is my Part 2 of exploring the Muse file system.
original post: https://x.com/heypeterjames/status/2103545183400800746
pete at mouse dot dev
Are you sure this isn't just a custom model that is API-compatible with OpenAI?
a fine-tuned OpenAI model on Azure for the purpose of compaction or something could make sense I guess but that would still be OpenAI's weights with meta ontop, and they already have a compaction model served under azure/avocado-compaction-v1
Source 2: I work on AI at Meta.
The Muse public APIs seem to be heavily inspired by OpenAI's, they even support the richer "responses" API. They have a similarly shaped compaction API, they have the same "encrypted" reasoning. Even the Muse harness seems heavily inspired by (open source) Codex. Considering Azure's historic involvement w/ OpenAI, it seems more plausible that Azure is used as spill-over capacity, and the same Azure infra used to encrypt GPT models encrypts Muse models.
My guess that the generally visible Anthropic/OpenAI details are leftovers from the whole meta "move fast" behavior. There are a few rough edges on the product where it leaks internal codenames (eg. signing in w/ WhatsApp required consenting to using "hatch" on one screen), so I'd believe that this was a leftover on the VM from the prototyping stage, before the Muse models were ready, or because the dev's got to benchmark against different models.
???
Every single provider basically copy openai's api endpoint
https://api.openai.com/v1/chat/completions
https://router.huggingface.co/v1/chat/completions
https://integrate.api.nvidia.com/v1/chat/completions
https://generativelanguage.googleapis.com/v1beta/openai/chat...
https://api.x.ai/v1/chat/completions
https://api.deepseek.com/chat/completions
https://openrouter.ai/api/v1/chat/completions
https://opencode.ai/zen/v1/chat/completions
https://api.cerebras.ai/v1/chat/completions
Saying "Muse's public APIs seem heavily inspired by OpenAI" is like saying "a web server’s API seems heavily inspired by HTTP". OpenAI’s API format has become an industry standard.
definitely tracks with alexandr’s hand-wavey influencer turn.
https://dev.meta.ai/docs/overview
Source: I work on AI at Meta.
How do you know?
Every cloud provider (AWS, Azure etc) is struggling with meeting LLM demand.
Source: first hand info
Because there was an article written in Bloomberg about how they're looking to sell their extra compute [0]
[0] https://archive.is/4ODV4