Muse Spark 1.3: price, plans and where it ranks
Official sources only.
Strengths and limits
- Meant for
- Meta's Muse Spark model (the company's own words)
- Best at
- coding (#8 on LiveBench 2.0 Coding); writing and chat (people's votes) (#10 on Arena Text (overall))
- Relatively weaker at
- reasoning (#12 on LiveBench 2.0 Reasoning)
- Context (how much it can read at once)
- 1M tokens
- Inputs and outputs
- Not published on the page we checked
- Effort / thinking levels
- Not published on the page we checked
- In which plans
- The company's plan pages don't name this model.
- Watch out for
- Cheaper Contributor mode lets Meta train on your data; Standard mode doesn't.
"Best at" and "weaker at" come from this model's best and worst placings on the independent leaderboards we carry (all rankings); we don't run our own tests. Other facts3. When to use which model.
API price
| Input / 1M | Output / 1M | Cached input / 1M | Notes | Source |
|---|---|---|---|---|
| US$1.25 | US$4.25 | US$0.15 | Standard mode price (your content is not used for training). Contributor mode for Spark 1.3/1.2: $0.10 in / $0.20 out / $0.002 cached, but in Contributor mode Meta can train its future models on your data (prompts, code, files and the responses). Web search grounding $2.50 per 1,000 queries | 3 |
Where Muse Spark 1.3 ranks
| Leaderboard | Rank | Score | Setting |
|---|---|---|---|
| Arena Text (overall)1 | #10 | 1496±5 | Max |
| LiveBench 2.0 Reasoning2 | #12 | 89.7 | – |
| LiveBench 2.0 Coding2 | #8 | 90.2 | – |
Different tests, different scales: never compare scores across rows. All rankings explained.
Sources checked Oct 26 to Nov 2, 2026.