Meta · Released Oct 15, 2026Muse Spark Open
Muse Spark Open is an 816B total / 42B active open-weight multimodal reasoning model from Meta for coding and long-horizon agentic work, with 1M-token context.
- Capability
- 155.1ECI · #39 of 165
- Input
- $1.25per 1M tokens
- Output
- $4.25per 1M tokens
- Context
- 1.05M131K max output
What it costs
- Input
- $1.25per million tokens
- Output
- $4.25per million tokens
- Blended (3:1)
- $2.00pricier than 72% of priced models
Typical monthly bills
| Monthly usage | Estimated cost |
|---|---|
| Side project2M input + 0.5M output tokens | $4.63 |
| Team assistant25M input + 5M output tokens | $52.50 |
| Production app250M input + 50M output tokens | $525.00 |
Median of 16 paid API providers on models.dev.
Independent benchmarks
#39 of 165 scored models
The shaded band is Epoch AI’s confidence range (153.1–157.3); the tick marks the median scored model.
Scores from Epoch AI, run independently of Meta.
The details
- Lab
- Meta
- Released
- Oct 15, 2026
- Knowledge cutoff
- Jan 4, 2026
- Context window
- 1,048,576 tokens
- Max output
- 131,072 tokens
- Inputs
- Text, Images, Audio, Video
- Output
- Text
- Reasoning
- Always on
- Tool calling
- Yes
- Structured output
- No
- Availability
- 16 API providerslisted on models.dev
What Meta claims
Published by the lab at launch. Settings vary, so compare these only with care.
| Benchmark | Score | Setting | Source |
|---|---|---|---|
| Artificial Analysis Intelligence Index | 39 | xhigh effort | Source |
| JobBench | 61.8 | xhigh effort | Source |
| MCP Atlas | 89.2 | xhigh effort | Source |
| GDPval-AA v2 | 1401 Elo | xhigh effort | Source |
| OSWorld-Verified | 80.4 | xhigh effort | Source |
| Toolathlon-Verified | 72.8 | xhigh effort | Source |
| HLE-Diamond | 26.1 | xhigh effort, no tools | Source |
| AA-Omniscience | 25 | xhigh effort | Source |
| SciCode | 58.5 | xhigh effort | Source |
| Humanity's Last Exam | 61 | xhigh effort, with tools | Source |
| GPQA Diamond | 89 | xhigh effort | Source |
| DeepSWE v1.1 | 60 | xhigh effort | Source |
More from Meta
About Muse Spark Open
How much does Muse Spark Open cost?
Muse Spark Open costs $1.25 input / $4.25 output per million tokens (median across 16 API providers). At a 3:1 input-to-output mix that is $2.00 per million tokens, more expensive than 72% of the 445 priced models we track.
What is the context window of Muse Spark Open?
Muse Spark Open accepts up to 1,048,576 tokens per request and can write up to 131,072 tokens in one response.
How good is Muse Spark Open?
Epoch AI gives Muse Spark Open a Capabilities Index score of 155.1 (likely range 153.1–157.3), ranking it #39 of 165 models Epoch has scored. Epoch AI benchmark results: GPQA Diamond 88.6%, FrontierMath Tiers 1–3 69.8%, OTIS Mock AIME 2024–2025 98.1%, SimpleQA Verified 56.6%.
Is Muse Spark Open open source?
Yes. Meta publishes the weights under the Muse Model License licence, so it can be downloaded and self-hosted.
What inputs does Muse Spark Open support?
Muse Spark Open accepts text, images, audio and video and replies in text. It is a reasoning model, supports tool calling.
When was Muse Spark Open released?
Meta released Muse Spark Open on Oct 15, 2026. Its training data runs to Jan 4, 2026.
What are the best alternatives to Muse Spark Open?
The closest current models from other labs on capability, price and release date are GLM-5.3 (Z.ai (Zhipu), ECI 156.2, $1.40 / $4.40), Grok 4.7 (xAI, ECI 154.6, $2.00 / $6.00), Gemini 3.6 Flash (Google, ECI 154.7, $0.75 / $3.75) and Qwen3.8 Max 0902 (Alibaba (Qwen), ECI 156.2, $2.00 / $6.00).