Meta released Muse Spark on April 10, its first major model update since early 2025. On LMSYS Chatbot Arena, Muse Spark exceeded expectations: 3rd in Text Arena (tied with Gemini 3.1 Pro and Claude Opus 4.6), 2nd in Vision Arena (tied with Claude Opus 4.6).
This marks Meta’s return after over a year of silence. Muse Spark also scored #4 in image generation quality on LMSYS. Combined with Meta’s open-source strategy on Llama, Muse Spark’s closed-source launch signals a shift from “pure open source” to a dual “open source + closed frontier” approach.
Current Leaderboard Overview
LMSYS Chatbot Arena (as of mid-April 2026):
| Rank | Model | Elo Score | Trend |
|---|---|---|---|
| 1 | Gemini 3.1 Pro | 1287 | ↑ |
| 2 | Claude Opus 4.6 | 1265 | → |
| 3 | GPT-5.3 | 1248 | → |
| 3 | Muse Spark | ~1248 | New |
Note: Opus 4.7 and GPT-5.5 may not yet be fully reflected in LMSYS rankings.
Artificial Analysis Intelligence Index: Claude Opus 4.7, GPT-5.4, and Gemini 3.1 Pro are tied in the top tier. Opus 4.7 scored 57, up 4 points from Opus 4.6.
Landscape Assessment
Muse Spark’s return expands frontier competition from a “big three” to a “big four.” Meta’s advantages: the world’s largest social data pipeline and open-source ecosystem foundation. If Muse Spark’s API pricing is competitive, it could directly challenge Gemini 3.1 Pro’s mid-range market share.
The “crowding” at the top is notable. The Elo gap between the top three has narrowed to within 40 points (1287 vs 1248), meaning perceived differences in daily use are shrinking. When performance gaps narrow, price, ecosystem, and developer experience replace “who has the highest score” as deciding factors.
Action Items
- Model selection reference: The top three on LMSYS Text Arena show minimal differences for general conversation. Filter by price, context window, and specific capabilities.
- Watch for Opus 4.7 and GPT-5.5 rankings: May shift significantly once more voting data accumulates.