Muse Wonder 1.0 Review: Meta's 1M-Token Agent Model With Text, Image, Video and Audio In
Pick Muse Wonder 1.0 for long agent runs and mixed-media prompts if $4 in and $20 out suits you; tool_choice is auto only, and this is not Wonder Full.
Freedom Score
Not yet tested
- Price
- $8.00 per 1M mixed tokens (3 input : 1 output)
- Output speed
- 120 tok/s
- Success rate
- Collecting data
Muse Wonder 1.0 is the standard Muse Wonder model from Meta, which Meta presents as a frontier model for coding, long-running agents and computer use. The model is labelled preview. On our API it is one OpenAI-compatible chat endpoint that takes text, images, video and audio in one message list, reads up to 1,048,576 tokens of input and writes back as many as 1,048,576, at $4 per million input tokens and $20 output.
- SpicyAPI Engineering
- Tested Oct 28, 2026
- Methodology v3.3 →
- We sell these models; scores ignore price →
The verdict
This is a catalogue review. We have not run our own quality or content tests, and we quote no benchmark scores: Meta's claims are for the larger Wonder Full variant, which is a separate subscription product and is not what this endpoint serves. Independent write-ups put other models ahead on some programming and terminal tests.
The case for it is the combination of window, input types and tool use. The case against is the price, which is steep next to the other models we added this week, and a tool-control limitation described below.
Key facts
- Tested by us: catalogue check and launch smoke run only, no quality or content tests; Meta's claims concern a different variant
- Price: $4 per million input tokens, $0.20 cached and $20 output
- Limits: 1,048,576-token context, replies up to 1,048,576 tokens (default 4,096), reasoning counts toward max_tokens, up to 128 tools
- Input: text, image, video, audio (wav, mp3)
- Reasoning: minimal, low, medium, high, xhigh or max; high is the default
- Sources: the scoring rules are on our review methodology page, and the full parameter list is on the Muse Wonder 1.0 model page.
What we liked and what held it back
What we liked
- A million-token window. Whole repositories or long agent transcripts in one request.
- Mixed media in one call. Image, video and audio parts alongside text.
- Six reasoning levels. From minimal to max.
- Parallel tool calls. Up to 128 tool definitions per request.
- JSON output with
response_formatand SSE streaming.
What held it back
- tool_choice accepts auto only. You cannot force a specific tool or require one; for plain text, do not send tools.
- High price. $4 and $20 per million tokens.
- Reasoning counts toward max_tokens and bills as output; a long reply plus deep thinking has to fit in the cap you set.
- Preview. Meta gives no release date, maximum output or knowledge cutoff.
- Not Wonder Full. The larger variant is not on this endpoint.
What we tested
We have not run Muse Wonder 1.0 through our capability sets or content levels. This page uses Meta's developer page for the model, the limits and parameters of our endpoint and the live price on the model page. We quote no benchmark numbers.
Prompts that worked
Starting points based on the endpoint's shape, not prompts we benchmarked.
1. Plan a library upgrade. A long-horizon planning task, the kind Meta positions the model for.
Plan a database library upgrade and identify compatibility checks before deployment.Run this prompt on Muse Wonder 1.0 →
2. A tool step. One function with tool_choice set to auto, the value this endpoint accepts.
Call report_status with status ready.Run this prompt on Muse Wonder 1.0 →
Settings that matter: start at medium reasoning and move to xhigh or max only for hard debugging and planning; label which attachment a question is about when you send several media parts; leave tool_choice at auto and omit tools when you want plain text.
Roleplay and how far it writes
We have not tested Muse Wonder 1.0 on our content levels yet, so this review makes no claim about how far it writes or how often it softens or refuses. The model page carries no content-test results for it either. On our API we add no content filter; whatever comes back is the model's own behaviour.
Not tested yet: all five content levels, roleplay quality over many turns, and languages other than English. We will add results here once the runs are done.
Every character in a request should be an adult, keep real, identifiable people out of it, and you're responsible for the law where you publish.
Alternatives
- Much cheaper, long window? Claude Haiku 5.5 reads 1,000,000 tokens of text and images and has a full review.
- A stronger Claude for hard work? Claude Sonnet 5.5 is a measured option in the same price band.
- Open weights, text only? Beam 1.0 has a similar window at a quarter of the input price.
How Muse Wonder 1.0 earns its Spicy Index
Each dimension below links to its full workings: the table row every point came from and why points were lost.
- Public leaderboards 40% of quality
- Not yet tested
- Role-play 15% of quality
- Not yet tested
- Creative writing 15% of quality
- 69.0
- Instruction following 15% of quality
- 99.0
- Long-context memory 15% of quality
- Not yet tested
- Capability 10.5% weight
- 64.0
- Freedom Score 30% weight
- Not yet tested
No exact match on a tracked public leaderboard yet. How it works
Our role-play suite hasn’t run on this model yet. How it works
Our long-context suite hasn’t run on this model yet. How it works
Freedom evidence is untested or insufficient. The Index retains the original quality and capability score, is marked preliminary and still receives a rank. Missing evidence is not zero. How it works
Freedom evidence is untested or insufficient. The Index retains the original quality and capability score, is marked preliminary and still receives a rank. Missing evidence is not zero.
Open the full test report →Capability, item by item
Points from the public scoring table. Upscaled tiers and limits we set ourselves don’t count.
- Context window (creator-published)Not declared
- 0.0 / 30
- Input modalities on SpicyAPIText, plus images, audio
- 19.0 / 25
- Tool useTools and tool_choice
- 15.0 / 15
- Structured outputAccepted, schema adherence verified
- 15.0 / 15
- Reasoning control6 levels: minimal · low · medium · high · xhigh · max
- 15.0 / 15
What Muse Wonder 1.0 can do on SpicyAPI
Straight from the live model specification on SpicyAPI. Some creators offer more in their own apps; this is what you can call through our API today.
| Feature | Chat |
|---|---|
| Context window | 1,048,576 tokens |
| Max output per request | 1,048,576 tokens |
| Reasoning levels | minimal · low · medium · high · xhigh · max |
| Tool calling | Yes |
| Structured output | Yes |
How long it really takes
Measured on real jobs over the last 30 days, from request to finished result. Times include our own test runs.
- Output speed
- 120 tok/sIncludes time to first token
- Success rate
- Collecting data
A time appears once at least 10 completed jobs are in. Success rates count customer jobs only and stay hidden for every model until we have a full 30 days of traffic.
Models to consider instead of Muse Wonder 1.0
Anthropic
Claude Haiku 5.5
Much cheaper, long window?
- Spicy Index
- 76.4−4.6
- Freedom Score
- 77.8
- Price
- $0.20−98%/ 1M tokens · 3:1 input/output
Anthropic
Claude Sonnet 5.5
A stronger Claude for hard work?
- Spicy Index
- 79.6−1.4
- Freedom Score
- 88.3
- Price
- $4.00−50%/ 1M tokens · 3:1 input/output
Reflection
Beam 1.0
Open weights, text only?
- Spicy Index
- 68.9−12.1
- Freedom Score
- —
- Price
- $1.50−81%/ 1M tokens · 3:1 input/output
What do you think?
Was this review useful?
Log in to vote.
Comments
Comments are checked automatically and by our team. We remove spam, offensive content, personal attacks and anything that names a real person without consent. Report a comment and we review it within one business day.
Tried this model? Share what worked. You need a free account to comment, so we can keep spam out.
Muse Wonder 1.0 FAQ
Try Muse Wonder 1.0 yourself
Set every parameter in the Playground and see the exact price before you run it.
$8.00 per 1M mixed tokens (3 input : 1 output)
More reviews
See the leaderboards →GPT 6.1 Terra
Pick GPT 6.1 Terra for high-volume screens, documents and writing where reasoning_effort none helps; so far we have run only launch checks, no scored tests.
- Spicy Index
- 79.9
- Freedom Score
- —
Tested Oct 28, 2026
Read the reviewKimi K3.1 Pro
Pick Kimi K3.1 Pro for long-context work with screenshots and clips when you accept always-on thinking; so far we have run only launch checks, no scored tests.
- Spicy Index
- 80.4
- Freedom Score
- —
Tested Oct 28, 2026
Read the review
Grok 4.3
Pick Grok 4.3 for a huge context window, a genuine reasoning off switch and uncensored writing at a low price; skip it if you need file uploads.
Tested Sep 24, 2026
Read the review