Muse Spark 1.3
Meta · released Sep 2, 2026
Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It is designed to keep track of information across extended tasks, work through...
Specification
- Context window
- 1.05M
- Max output
- 944K
- Knowledge cutoff
- Not stated
- Parameters
- Undisclosed
- Licence
- Proprietary
- Serving providers
- 1
- Moderated
- Yes
- Uptime
- 100.0%
Intelligence
48.1
97th percentile
Coding
75.8
Coding Index
Agentic
55.5
Agentic Index
Output speed
65 t/s
Median across providers
Latency
3.81s
Time to first token
Cost per task
$0.40
Estimated
Benchmarks
Where the score comes from
The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.
Evaluation scores
Percentage correct · higher is better
- GPQA Diamond93.5%
- AA-LCR (long context)83.0%
- SciCode58.8%
- τ²-bench (Telecom)50.5%
- Humanity's Last Exam48.7%
An evaluation missing from this list was not run for this model — it is not a zero.
View as table
| Evaluation | Score |
|---|---|
| GPQA Diamond | 93.5% |
| AA-LCR (long context) | 83.0% |
| SciCode | 58.8% |
| τ²-bench (Telecom) | 50.5% |
| Humanity's Last Exam | 48.7% |
Against its peers
Intelligence Index · this model highlighted, nearest peers in grey
- Claude Opus 5Anthropic50.8
- Claude Fable 5Anthropic49.6
- Muse Spark 1.3Meta48.1
- GPT-6 SolOpenAI47.6
- GPT-5.6 SolOpenAI47.0
- Grok 4.7xAI46.4
- MiMo-V2.6-ProXiaomi46.3
- 45.4
Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.
View as table
| Model | Intelligence |
|---|---|
| Claude Opus 5 | 50.8 |
| Claude Fable 5 | 49.6 |
| Muse Spark 1.3 | 48.1 |
| GPT-6 Sol | 47.6 |
| GPT-5.6 Sol | 47.0 |
| Grok 4.7 | 46.4 |
| MiMo-V2.6-Pro | 46.3 |
| Qwen3.8 Max (0902) | 45.4 |
Percentile among all indexed models
Pricing
What it costs to run
List prices per million tokens, plus what one representative task works out to.
List price
- Input / 1M tokens
- $1.25
- Output / 1M tokens
- $4.25
- Cached input / 1M
- $0.15
- Blended 3:1
- $2
One task, estimated
$0.40
- Input tokens
- 50,000
- Output tokens
- 80,000
- Profile
- Reasoning
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).
Serving providers
Speed and latency figures are medians across these providers, so a widely-served model reports a blend rather than any single endpoint.
Arena
Head-to-head generation quality
Elo from pairwise judgements, broken out by the kind of thing the model was asked to build.
- Overall Elo
- 1,360
- Win rate
- 58.9%
- Strongest at
- 3D scenes
- Tournaments
- 5,448
Elo by category
Dot position on a 1,200–1,375 scale · Elo has no meaningful zero
- 1,360
- 1,358
- 1,354
- 1,344
- 1,332
- 1,301
- 1,295
- 1,219
Agent categories (full-stack apps, mobile apps) are only scored for models tested in agent harnesses.
View as table
| Category | Elo | Win rate |
|---|---|---|
| 3D scenes | 1360 | 58.9% |
| Websites | 1358 | 58.2% |
| UI components | 1354 | 57.8% |
| Data visualisation | 1344 | 58.1% |
| Game development | 1332 | 54.3% |
| SVG | 1301 | 55.3% |
| Full-stack apps | 1295 | 56.2% |
| Mobile apps | 1219 | 47.9% |