Meta · released Oct 15, 2026
Muse Spark Open
Muse Spark Open is Meta's new open-weight coding model, and early results put it mid-pack.
NewOpen weights2 weeks old
- Input price
- $1.25 / M
- Output price
- $4.25 / M
- Context
- 1M tokens
- Longest reply
- –
- Reads
- text, image, video, file, audio
- Knows up to
- –
Where it shines
- Its weights are free and run at home on two M5 Ultra Mac Studios.
- It scores 56.6 on SimpleQA, high for an open model.
- It holds 1M tokens of context.
Where it struggles
- Its web dev score of 1526 trails other open models.
- No consumer chat app offers it yet.
Where you can use it
- Hugging FaceDownloadWeights also on major clouds.
- OpenRouterAPI$1.25 in / $4.25 out. Groq, Together and Fireworks also host it.
- OpenRouterAPIListed as
meta/muse-spark-open
How it scores on each job
Scores run 0 to 100 and blend every measure we have for that job. The number after the dot is its rank among the 33 models in the lineup.
The evidence
| Measure | Score | Rank in lineup | Source |
|---|---|---|---|
| Epoch Capabilities Index | 155 | #22 | Epoch AI |
| WebDev Arena | 1526 | #24 | LMArena |
| GPQA Diamond | 89% | #22 | Epoch AI |
| FrontierMath | 70% | #17 | Epoch AI |
| SimpleQA Verified | 57% | #11 | Epoch AI |
| ARC-AGI-2 | 58% | #20 | Epoch AI |
Compare it with these
8 weeks old
Meta
Muse Spark 1.3
Muse Spark 1.3 is Meta's mid-price model and reads documents well for its cost.
Long documentsEveryday questions
- Price
- $1.25 in · $4.25 out
- Context
- 1M
Open weights3 months old
Z.ai
GLM-5.3
GLM-5.3 is Z.ai's open-weight model and handles chat decently at a low price.
Everyday questions
- Price
- $1.2 in · $4.18 out
- Context
- 1M
8 months old
Gemini 3.1 Pro Preview
Gemini 3.1 Pro Preview is Google's older flagship, and Gemini 4 Argon beats it for less money.
Writing
- Price
- $2 in · $12 out
- Context
- 1M