Meta's Muse Wonder lands on Vercel's AI Gateway for coding and agents

What Muse Wonder is built for

Muse Wonder is a model from Meta that is designed for coding, long-running agents, and computer use. According to the Vercel changelog, it can plan multi-step tasks, delegate work to sub-agents, and recover from errors as work progresses.

The model has a 1M-token context window and supports text, image, and PDF input. It also supports tool calling and structured output. Vercel says this lets agents work with code, screenshots, and documents in the same workflow.

How to access it through AI Gateway

Developers can call the model through AI Gateway with one API key, without creating a separate Meta account. The setup is to set AI_GATEWAY_API_KEY in the environment and then use meta/muse-wonder-1.0 as the model name.

Vercel's example uses the streamText function from the ai package. It passes the model name and a prompt, "Plan a codebase refactor and identify the tools needed.", then writes the streamed text to standard output.

For coding agents, Vercel suggests running npx vercel ai-gateway setup and selecting meta/muse-wonder-1.0 in the agent's model settings. The model can also be tried in the model playground, and Vercel points to its model catalog for further options.

What AI Gateway offers and what it means for developers

AI Gateway is Vercel's single API for calling models. According to Vercel, it also covers tracking usage and cost, and configuring routing, retries, and failover. Users can authenticate with an AI Gateway API key or bring their own provider key.

For teams building agents, the addition means Muse Wonder can be tried alongside other models in the catalog using the same key and the same usage and cost tracking. The source does not give pricing or performance figures for the model, so those details are not covered here.