Meta · Muse Spark
Muse Spark 1.3
Meta Superintelligence Labs' lighter closed model, now a tier below Muse Wonder, with a 1M context window and multimodal input.
- AA Intelligence
- 48.1#11 of 151
- Input / output per 1M
- $1.25/ $4.25
- Output speed
- 227tok/s
- Context window
- 1.05M tokens
Composite indexes
- Artificial Analysis Intelligence Index#11 of 15148
Leader: Claude Opus 5.5 at 58
Coding
- Terminal-Bench 4.0#18 of 9633.3%
Leader: Claude Sonnet 5.5 at 70.6%
- SciCode#11 of 10058.8%
Leader: Claude Opus 5.5 at 66.9%
Agents and tool use
- GDPval-AA#7 of 1091674
Leader: Muse Wonder Full at 1857
Reasoning and knowledge
- Humanity’s Last Exam#11 of 15348.7%
Leader: Claude Sonnet 5.5 at 64.5%
- GPQA Diamond#11 of 14493.5%
Leader: GPT-6 Astra at 96.1%
- CritPt#12 of 14424.9%
Leader: Muse Wonder Full at 33%
Vision and long context
- AA Long Context Reasoning#14 of 14483%
Leader: Kimi K3 at 88.7%
Human preference
- LMArena Text#5 of 571493
Leader: Gemini 4 Argon at 1525
- LMArena WebDev#10 of 581658
Leader: Claude Opus 5.5 at 1818
Other published results
| Result | Score | Reported by | Source |
|---|---|---|---|
| AA-Omniscience Index · max effort | 25 index (-100 to 100) | independent | Artificial Analysis: Muse Spark 1.3 (max) ↗ |
| AutomationBench-AA (partial score) · max effort | 57.9% | independent | Artificial Analysis: Muse Spark 1.3 (max) ↗ |
| Terminal-Bench Science (AA) · max effort | 11% | independent | Artificial Analysis: Muse Spark 1.3 (max) ↗ |
| AA Coding Agent Index v1.5 · Muse Code + Muse Spark 1.3 (max); DeepSWE 71.7 / SWE-Atlas-QnA 59.4 / TB4 31.8 | 54.3 index | independent | Artificial Analysis Coding Agents ↗ |
| AA Intelligence Index (xhigh) · AA Intelligence Index v4.3.2, xhigh | 45.1 index | independent | Artificial Analysis: Muse Spark 1.3 (xhigh) ↗ |
| LMArena Code (xhigh) · muse-spark-1.3 xHigh | 1626 Elo | independent | LMArena Code (WebDev) leaderboard ↗ |