Muse Wonder

Meta's most capable Muse model for agents, knowledge work and coding, with a 1M-⁠token context window

About Muse Wonder

Muse Wonder is Meta's most capable Muse model, built for agents, knowledge work and coding. It reads text, images, video, audio and PDFs, holds up to 1 million tokens in one conversation, and keeps working on a hard task by trying other approaches before it gives up

Muse Wonder scores 68.1% on AutomationBench, ahead of GPT-⁠6.1 Terra at 60.2%, and 30.6% on GDP.pdf, one of the ten best results of any model on questions about long professional PDFs. When it doesn't know the answer to a knowledge question, it says so instead of making one up in 79.1% of cases

On Overchat AI, Muse Wonder is in the same chat as Muse Spark, Claude and GPT-⁠6, so you can ask it, attach a file and switch models without starting over

Developer

Muse Wonder vs other models

Muse Wonder ranks #11 on the Artificial Analysis Intelligence Index out of 54 models

Artificial Analysis Intelligence Index44#11 of 54 modelsBetter than 79% of 54 models

#9of 148 modelsGDP.pdfQuestions about long professional PDFs with tables, charts and fine print30.6%
#11of 142 modelsSciCodePython code for scientific computing60.3%
#13of 111 modelsNot making facts upHow often the model says it doesn't know instead of making up an answer79.1%
#15of 150 modelsAutomationBenchBusiness workflows an agent runs across many apps68.1%

Source: Artificial Analysis

Compare Muse Wonder

Muse Wonder vs Kimi K3.1 Pro

Muse Wonder scores higher than Kimi K3.1 Pro in 3 of 7 benchmarks. The widest lead is in FrontierCode 2.0: 27.9% against 25.5%. Kimi K3.1 Pro is ahead in Terminal-Bench 4.0: 67.2% against 52.0%. Muse Wonder writes 980 tokens per second, Kimi K3.1 Pro writes 340

Benchmarks won: Muse Wonder: 3; Kimi K3.1 Pro: 4
Average score: Muse Wonder: 42.0%; Kimi K3.1 Pro: 45.5%
Output speed, tokens/s: Muse Wonder: 980; Kimi K3.1 Pro: 340
Time to first token, s (log scale): Muse Wonder: 8.7; Kimi K3.1 Pro: 7.4

Kimi K3.1 Pro aheadMuse Wonder ahead

FrontierCode 2.0+2.4
HLE-Diamond+1.7
CritPt v2 Hard+1.0
GDP.pdf−0.8
AutomationBench−5.2
Terminal-Bench-Science−8.6
Terminal-Bench 4.0−15.2

Muse Wonder vs GPT-6.1 Terra

Muse Wonder scores higher than GPT-⁠6.1 Terra in 2 of 7 benchmarks. The widest lead is in AutomationBench: 68.1% against 60.2%. GPT-⁠6.1 Terra is ahead in Terminal-Bench-Science: 57.6% against 25.7%. Muse Wonder writes 980 tokens per second, GPT-⁠6.1 Terra writes 75

Benchmarks won: Muse Wonder: 2; GPT-6.1 Terra: 5
Average score: Muse Wonder: 42.0%; GPT-6.1 Terra: 51.4%
Output speed, tokens/s: Muse Wonder: 980; GPT-6.1 Terra: 75
Time to first token, s (log scale): Muse Wonder: 8.7; GPT-6.1 Terra: 4.3

GPT-6.1 Terra aheadMuse Wonder ahead

AutomationBench+7.9
GDP.pdf+2.0
Terminal-Bench 4.0−3.6
HLE-Diamond−8.3
FrontierCode 2.0−14.4
CritPt v2 Hard−17.5
Terminal-Bench-Science−31.9

Muse Wonder vs Grok 4.7

Muse Wonder scores higher than Grok 4.7 in 6 of 7 benchmarks. The widest lead is in Terminal-Bench 4.0: 52.0% against 25.8%. Grok 4.7 is ahead in FrontierCode 2.0: 28.7% against 27.9%. Muse Wonder writes 980 tokens per second, Grok 4.7 writes 81

Benchmarks won: Muse Wonder: 6; Grok 4.7: 1
Average score: Muse Wonder: 42.0%; Grok 4.7: 29.7%
Output speed, tokens/s: Muse Wonder: 980; Grok 4.7: 81
Time to first token, s (log scale): Muse Wonder: 8.7; Grok 4.7: 8.9

Grok 4.7 aheadMuse Wonder ahead

Terminal-Bench 4.0+26.2
CritPt v2 Hard+17.0
HLE-Diamond+15.8
Terminal-Bench-Science+14.3
GDP.pdf+10.6
AutomationBench+2.5
FrontierCode 2.0−0.8

Muse Wonder vs Muse Spark

Muse Wonder scores higher than Muse Spark in 7 of 7 benchmarks. The widest lead is in Terminal-Bench 4.0: 52.0% against 33.3%. Muse Wonder writes 980 tokens per second, Muse Spark writes 1129

Benchmarks won: Muse Wonder: 7; Muse Spark: 0
Average score: Muse Wonder: 42.0%; Muse Spark: 33.4%
Output speed, tokens/s: Muse Wonder: 980; Muse Spark: 1129
Time to first token, s (log scale): Muse Wonder: 8.7; Muse Spark: 6.2

Muse Spark aheadMuse Wonder ahead

Terminal-Bench 4.0+18.7
Terminal-Bench-Science+14.7
AutomationBench+10.2
HLE-Diamond+5.3
FrontierCode 2.0+4.8
GDP.pdf+4.0
CritPt v2 Hard+2.5

Best model by use case

Average benchmark scores of Muse Wonder and 4 other models by use case

Coding
?Average of 3 tests
1. 1GPT-6.1 Terra51.8%
2. 2Kimi K3.1 Pro42.3%
3. 3Muse Wonder35.2%
4. 4Muse Spark22.5%
5. 5Grok 4.722.0%
Knowledge work
?Average of 2 tests
1. 1Kimi K3.1 Pro52.3%
2. 2Muse Wonder49.3%
3. 3GPT-6.1 Terra44.4%
4. 4Grok 4.742.8%
5. 5Muse Spark42.3%
Reasoning
?Average of 2 tests
1. 1GPT-6.1 Terra57.6%
2. 2Muse Wonder44.8%
3. 3Kimi K3.1 Pro43.4%
4. 4Muse Spark40.9%
5. 5Grok 4.728.4%

Muse Wonder benchmarks

Compare Muse Wonder with 4 other models in 7 benchmarks measuring coding, knowledge work and reasoning

Order

Coding

Terminal-Bench 4.0?52.0%−15.2
FrontierCode 2.0?27.9%−14.4
Terminal-Bench-Science?25.7%−31.9

Knowledge work

AutomationBench?68.1%−5.2
GDP.pdf?30.6%−0.8

Reasoning

HLE-Diamond?39.0%−8.3
CritPt v2 Hard?50.5%−17.5

Sources: Meta, Introducing Muse Wonder (Oct 12, 2026), Artificial Analysis, Muse Wonder and Muse Wonder Full

Muse Wonder speed

Output speed is how fast the model writes (tokens per second, higher is better). Time to first token is how long you wait before the answer starts (lower is better)

Output speed and time to first token of 35 models

?Percentile

Time to first token, s (log scale)

Output speed, tokens/s (log scale)

Output speed and time to first token of 35 models
Models	Output speed, tokens/s (log scale)	Time to first token, s (log scale)
GPT-6 Luna	143	2.4
GPT-6.1 Terra	75	4.3
GPT-6.1 Sol	58	2.9
GPT-6 Astra	48	3.2
Claude Opus 5.5	88	1.9
Claude Opus 5	54	2.6
Claude Opus 4.8	59	1.3
Claude Fable 5.2	84	2.3
Claude Sonnet 5.5	102	1
Claude 5 Sonnet	215	10.5
Claude Sonnet 4.6	220	10.8
Claude Haiku 5.5	167	2.1
Gemini 4 Carbon	169	29.5
Gemini 4 Argon	91	6
Gemini 3.8 Flash	203	7.5
Gemini 3 Pro	270	11.2
Grok 4.7	81	8.9
Grok 4.6	290	32
Qwen 3.8 Max	81	30.6
DeepSeek V4 Pro	144	1.1
DeepSeek V4.1 Flash	422	2.2
Mistral Large 4	226	8.6
Hy4	214	37.7
ERNIE 5.5 Flash	405	2.7
Muse Wonder	980	8.7
Muse Spark	1129	6.2
Kimi K3.1 Pro	340	7.4
Kimi K3	338	5.9
GLM 5.3	197	1.2
Flint	194	12.3
Llama 4	81	0.7
Claude 4.5 Haiku	162	0.9
MiniMax M3	169	4.5
MiniMax M3.1 Flash	656	7.2
Kimi K2.7 Code	455	3.6

Typing test

Every model writes the same answer. The test waits the model's time to first token, then prints the text at the model's output speed and shows the total time

Answer length

Muse Wonder980 tokens/s8.7 s to first token
Kimi K3.1 Pro340 tokens/s7.4 s to first token
GPT-6.1 Terra75 tokens/s4.3 s to first token
Grok 4.781 tokens/s8.9 s to first token

Source: Overchat AI, 2026-11-11

See how Muse Wonder answers

Examples of Muse Wonder answers to 3 prompts: Find what's wrong, Simulate a pendulum and Check a fact. Each answer is the unedited output of the model

My online store's last six months: visits 42k, 45k, 47k, 51k, 50k, 53k; orders 1,260, 1,280, 1,240, 1,230, 1,150, 1,140; refunds 38, 41, 52, 66, 71, 80. Find the two biggest problems and suggest one fix for each, in a short table.

Muse Wonder

Problem	Evidence (Month 1 → 6)	Fix
1. Falling conversion rate	Visits up 42k → 53k (+26%), but orders down 1,260 → 1,140. Conversion fell 3.0% → 2.15%.	Audit the new traffic source and align its landing page to the ad/search intent.
2. Rising refund rate	Refunds up 38 → 80; refund rate more than doubled, 3.0% → 7.0% of orders.	Identify the #1 refund reason for the top product and fix it (e.g., sizing/description or quality).

Write a Python function that simulates a damped pendulum with the RK4 method and returns the angle over time. Keep it under 40 lines and explain your choice of time step in one sentence.

Muse Wonder

python

import numpy as np

def simulate_damped_pendulum(theta0=1.0, omega0=0.0, t_max=10.0,
                             dt=0.01, g=9.81, L=1.0, damping=0.1):
    n = int(t_max / dt) + 1
    t = np.linspace(0, t_max, n)
    theta = np.empty(n)
    omega = np.empty(n)
    theta[0], omega[0] = theta0, omega0

    def deriv(th, om):
        return om, -(g / L) * np.sin(th) - damping * om

    for i in range(n - 1):
        th, om = theta[i], omega[i]
        k1_th, k1_om = deriv(th, om)
        k2_th, k2_om = deriv(th + 0.5 * dt * k1_th, om + 0.5 * dt * k1_om)
        k3_th, k3_om = deriv(th + 0.5 * dt * k2_th, om + 0.5 * dt * k2_om)
        k4_th, k4_om = deriv(th + dt * k3_th, om + dt * k3_om)
        theta[i + 1] = th + (dt / 6) * (k1_th + 2 * k2_th + 2 * k3_th + k4_th)
        omega[i + 1] = om + (dt / 6) * (k1_om + 2 * k2_om + 2 * k3_om + k4_om)

    return t, theta

I chose dt=0.01 s because it is far smaller than the pendulum's ~2 s natural period, so RK4 remains accurate and stable without unnecessary computation.

Who was the first person to walk on the Moon, and who was the 13th? If any part of the question has no real answer, say so instead of guessing.

Muse Wonder

The first person to walk on the Moon was Neil Armstrong of Apollo 11, on July 20, 1969 (EDT).

There is no 13th person — only 12 people in total have walked on the Moon, all Apollo astronauts between 1969 and 1972.

More prompts to try with Muse Wonder

What Muse Wonder is best for

Agents and business workflows
Muse Wonder runs workflows across sales, support and finance apps, with 68.1% on AutomationBench, ahead of GPT-⁠6.1 Terra and Grok 4.7
Reports and long PDFs
Muse Wonder answers questions about long professional PDFs with tables, charts and fine print, with 30.6% on GDP.pdf, one of the ten best results of any model
Answers you can trust
Muse Wonder says when it doesn't know: in 79.1% of the knowledge questions it can't answer, it admits it instead of making up a fact
Coding and science
Muse Wonder writes scientific code, with 60.3% on SciCode, and completes hard terminal tasks, with 52.0% on Terminal-Bench 4.0

How to use Muse Wonder on Overchat AI

Start in the chat at the top of this page or in the Overchat AI app on the web, iOS or Android

01
Pick Muse Wonder
It is already picked in the chat on this page. In the Overchat AI app, open the model list next to the message field and pick Muse Wonder
02
Attach your code or a file
Add code, a document, a spreadsheet or a screenshot, and write what you need: the result, the format and what must stay the same
03
Keep going in the same chat
Ask a follow-up, ask for tests or a shorter version, or switch to another model without losing the conversation

Why use Muse Wonder on Overchat AI

One account for chat, images and video, on every device

Superchat picked over Muse Wonder: one request done by Claude Opus 5.5 (the plan) and Nano Banana 2.1 (the poster)

Superchat AI agent

Superchat Agent automatically routes your message to the model best suited for the task. It can generate images and videos, use the AI tools built into Overchat AI, and learn from your interactions to get better over time

Message field with a PDF, a spreadsheet and a dashboard screenshot attached, Muse Wonder picked and a voice message being recorded

Chat with images and files

Upload PDFs, spreadsheets, code and screenshots and ask Muse Wonder about them. You can also dictate a message instead of typing it

An image made with Nano Banana 2.1 and a surfing video made with Seedance 2.5

Go beyond text

Generate images with Nano Banana and GPT Image and videos with Seedance and Kling in the same account, with the same credits as chat

On the web and iPhone

Start a chat on your laptop and finish it on your phone. The Overchat AI app has the same models, and your chats stay in one account on every device

Meta models on Overchat AI

Compare 4 Meta models on Overchat AI by output speed, time to first token and the price of one message

Models

Output speed, tokens/s

First token, s

Credits per message?

Muse WonderThis page

980

8.7

Muse Spark

1129

6.2

Muse Ember

—

Llama 4

0.7

Source: Overchat AI, 2026-11-11

Explore other models

Models marked Soon are coming to Overchat AI; the rest are available now

More models

Learn more about AI models

All articles

We tested Gemini 4 Argon against Claude Opus 5.5, GPT-⁠6 Astra and Muse Wonder

We tested Gemini 4 Argon against Claude Opus 5.5, GPT-⁠6 Astra and Muse Wonder on three coding tasks. See benchmarks, speed and how to use it

Muse Wonder vs Muse Spark: Meta's New Model Tested

We tested Muse Wonder against Muse Spark 1.3 on logic, writing and coding tasks. See benchmarks and which Meta model is better

10 Best All-in-One AI Platforms in 2026: Why One AI Subscription Beats Many

Compare the 10 best all-in-one AI platforms in 2026, from Overchat AI to Poe and Magai, with prices, free plans and the newest GPT and Claude models

Muse Wonder FAQ

If you have any questions,
reach out to us at hello@overchat.ai

Updated November 11, 2026

Muse Wonder is Meta's most capable Muse model, released on October 12, 2026. It is built for agents, knowledge work and coding, reads text, images, video, audio and PDFs and holds up to 1 million tokens in one conversation. It carries important context across conversations and keeps working on hard tasks by trying other approaches

Muse Wonder Full is the stronger version of Muse Wonder, with 50 against 44 on the Artificial Analysis Intelligence Index. It is only in Meta's Muse Code subscription, while Muse Wonder is in the Meta Model API, OpenRouter and Overchat AI

Yes. Muse Wonder scores higher on every test it shares with Muse Spark 1.3: 52.0% on Terminal-Bench 4.0 against 33.3%, 68.1% on AutomationBench against 57.9% and 39.0% on HLE-Diamond against 33.7%. Muse Spark costs fewer credits per message on Overchat AI, so it is a good pick for simpler tasks

Muse Wonder is ahead at writing ready-to-merge code and on hard reasoning questions, with 27.9% on FrontierCode 2.0 against 25.5% and 39.0% on HLE-Diamond against 37.3%. Kimi K3.1 Pro is stronger at terminal tasks, with 67.2% on Terminal-Bench 4.0 against 52.0%. Both models are in Overchat AI

Muse Wonder runs multi-step agent tasks, answers questions about long reports and PDFs, writes and fixes code, and reads images, video and audio. On Overchat AI, attach a file or a screenshot and ask about it

Muse Wonder holds up to 1 million tokens in one conversation, about 750,000 words, and writes up to 943,718 tokens in one reply

Type your message in the chat at the top of this page, attach a file if you need to and press Send. After you sign in, Muse Wonder answers in the Overchat AI chat, where you can keep the conversation going on the web, iOS or Android and switch to another model at any time

More tools

AI Code Generator AI Text Summarizer AI Playground Claude Fable 5.2 GPT-6.1 Terra Gemini 4 Carbon Kimi K3.1 Pro Hy4

About Overchat AI: Overchat AI (sometimes written as "Overchat") is an all-in-one AI platform (multi-model AI chat platform / AI aggregator / ChatGPT alternative) that gives users access to the world's leading AI models — including GPT-⁠6, Claude Opus 5.5, Gemini 3.8, Grok 4.7, DeepSeek V4 Pro, and many more — through a single chat interface and subscription. Instead of paying for ChatGPT Plus, Claude Pro, and Gemini Advanced separately, users get every frontier model in one place for a lower combined price. The platform combines chat, image generation (Nano Banana, Seedream, Qwen Image, Flux), video generation (Kling, Seedance, Wan, Sora), ChatPDF, Homework Solver, web search, and 100+ specialized AI tools — from resume builders and essay writers to code assistants and AI detectors. Files, PDFs, images, and links can be attached to any conversation. Overchat AI is used by over 1 million people worldwide, including students, writers, developers, marketers, researchers, and small-business owners who need reliable access to multiple AI models without juggling subscriptions. It is available on the web at overchat.ai and on iOS

Muse Wonder

About Muse Wonder

Muse Wonder vs other models

Compare Muse Wonder

Muse Wonder vs Kimi K3.1 Pro

Muse Wonder vs GPT-6.1 Terra

Muse Wonder vs Grok 4.7

Muse Wonder vs Muse Spark