← Models
Muse Spark 1.3
Meta · proprietary · muse-spark-1.3
Muse Spark 1.3 is Meta Superintelligence Labs' proprietary multimodal reasoning model for long-running agentic, multi-agent, and coding workflows. It improves on Muse Spark 1.2 in long-horizon collaboration, multitasking, instruction following, tool use, failure recovery, and concise coding execution. It supports a 1,048,576-token context window and is available through Muse Code and the Meta Model API in standard and contributor tiers. ZeroEval routes to the lower-cost contributor endpoint, where prompts and outputs may be used to improve Meta's products, while standard pricing remains the default comparison basis.
Benchmark scores
| Benchmark | Score |
|---|---|
| GDPval-AA | 1754.00 |
| MRCR v2 (8-needle) | 0.98 |
| MRCR v2 (8-needle, 512K-1M) | 0.98 |
| DeepSearchQA | 0.89 |
| Terminal-Bench 2.1 | 0.89 |
| DeepSWE 1.1 | 0.75 |
Pricing
- Meta Model API$0.10 / $0.20
Input / output per 1M tokens
AA metrics
No Artificial Analysis link yet.
Arena Elo
- No Arena snapshot linked.
