← comparisons · head to head
Muse Wonder vs Gemini 4 Argon
Of the 6 public leaderboards that rate both, Muse Wonder scores higher on 1 and Gemini 4 Argon on 5. On the ArtificialWatch Index, which takes the median of each model's percentile across the boards that score it, Muse Wonder is at 68.7 and Gemini 4 Argon at 78.4 (out of 100).
Scores as published by each board; table composed November 13, 2026.
Get the next launch alert, free →
Leaderboard by leaderboard
| Board | Muse Wonder | Gemini 4 Argon | Higher |
|---|---|---|---|
| LMArena Elo (community votes) | 1,723.5 | 1,681 | Muse Wonder |
| LiveBench global average (0-100) | 70.7 | 77.7 | Gemini 4 Argon |
| LLM Stats LLM Stats Score · composite (0-100), their rendered top 15 | 52.7 | 53.6 | Gemini 4 Argon |
| Vals AI Vals Index accuracy % · weighted finance + coding tasks | 66.8 | 69.4 | Gemini 4 Argon |
| Artificial Analysis intelligence index · measured runs only | 50.1 | 50.4 | Gemini 4 Argon |
| Epoch AI Epoch Capabilities Index (ECI) · many benchmarks on one scale | 161.7 | 166.5 | Gemini 4 Argon |
Raw scores are each board's own scale, so compare within a row, not across rows. Full table: the bench index.
For coding
On the ArtificialWatch coding index, Muse Wonder is at 70 (#14 of 48) and Gemini 4 Argon at 95.3 (#2). Gemini 4 Argon scores higher on all 3 coding boards that rate both.
| Coding board | Muse Wonder | Gemini 4 Argon |
|---|---|---|
| LMArena · Coding | 1,515.6 | 1,540.2 |
| LiveBench · Coding | 74.9 | 85.7 |
| Vals Vibe Code Bench | 86 | 91.9 |
Muse Wonder at a glance
- Developer
- Meta
- Status
- Live since October 12, 2026
Live — Oct 12 · Meta's most capable model, a new line above Muse Spark · muse-wonder-1.0 on Meta Model API (preview) at $4 / $20 per 1M, $0.20 cached, no long-context premium · 1M ctx, 131K max output · Muse Wonder Full, the stronger variant, is in Muse Code only (Power and Maximum plans) · same $4 / $20 as Claude Opus 5.5 · Artificial Analysis Intelligence Index (read on AA Oct 12): Wonder 52, Wonder Full 56; Claude Opus 5.5 58 on the same index per Meta's chart · CursorBench 4.0 (read on cursor.com Oct 15): 50.4% on Max at $5.25 a task; on Cursor's main chart that trails Opus 5.5 (57.8%), Sonnet 5.5 (55.5%), Gemini 4 Argon (52.9%) and Fable 5.1 (51.8%), and Argon on its Low setting scores 50.6% at $4.09 · Meta's own table has Wonder Full ahead of Opus 5.5 on 8 of 14 rows (two are Meta-internal: its Personal Superintelligence Index and MemoryBench; DeepSWE 1.1 79.0 vs 74.2, SWE-Marathon 57.5 vs 46.3) and behind on the other 6, including Terminal-Bench 4.0, ProgramBench (59.0 vs 70.5) and the AA index · Meta's launch materials do not use the Watermelon codename, but @kimmonismus, who says he had early access, called the model Meta's 'Watermelon' on Oct 12 after launch, so it was very likely Wonder's codename (Meta has not said).
Meta's Muse gets voice, live avatars and Spaces, runs on models trained with Muse Wonder, and launches in Mexico
2026-11-06 · Meta shipped four updates to Muse, its personal agent app, on November 5 and 6. Muse now talks, shows a live avatar in the US, builds small apps called Spaces, and runs on specialized models trained with Muse Wonder, Meta's most capable model. Muse also launched in Mexico.
Meta ships Muse Wonder at Opus 5.5's $4 / $20, and says Wonder Full beats Opus 5.5 on 8 of 14 benchmarks
2026-10-12 · Meta released Muse Wonder today, its most capable model and a new line above Muse Spark. Muse Wonder is on Meta Model API now as muse-wonder-1.0, at $4 per million input tokens and $20 per million output, the same price as Claude Opus 5.5.
Gemini 4 Argon at a glance
- Developer
- Status
- Live since October 13, 2026
- API price
- $2 input / $10 output per 1M tokens · Google, gemini-4-argon
- Context window
- 1.05M tokens
Live — Oct 13 · GA as gemini-4-argon on the Gemini API (Google changelog), also in AI Studio, Vertex AI, the Gemini app and Antigravity · $2 / $10 per 1M through Nov 30, 2026, then $4 / $20 from Dec 1; cached input $0.10, then $0.20 (implicit caching only) · Batch and Flex half price, Priority $3.60 / $18 · on the API, Google's pricing page says it is not on the Free tier or on the Google AI Pro plan; the consumer Google AI Pro and Ultra subscriptions do include it in the Gemini app (gemini.google, read Oct 15) and in Antigravity's desktop app and CLI since Oct 13 (Google's Antigravity blog), where a screenshot @LuminaBench posted Oct 15 shows a separate weekly Argon limit · 1,048,576-token context shared by input and output; 65,536 max output by default, up to 1,048,576 with long decode · thinking low / medium / high · knowledge cutoff March 2026 (model card) · Google's own table: ahead of Claude Opus 5.5 on 14 of 18 shared rows, behind on FrontierSWE v2 (55.0 vs 62.3), Terminal-Bench 4.0 (57.4 vs 66.4), PostTrainBench and Terminal-Bench Science · Artificial Analysis Intelligence Index (read Oct 13): 52.6 at high, level with GPT-6 Astra (52.7), behind Opus 5.5 (57.6), Sonnet 5.5 (56.0), Muse Wonder Full (55.8) and Fable 5.1 (53.4) · #1 on the Vals Index (68.9%) · ARC Prize, verified (read Oct 15): ARC-AGI-3 97.5% through the provider adapter harness and 49.8% on the standard harness, ARC-AGI-2 91.3% ($0.32 a task), ARC-AGI-1 98.5% · Epoch Capabilities Index (read Oct 15): 164.6, fifth, behind Opus 5.5 (166.8), GPT-6 Astra (166.4), GPT-6.1 Sol (166.3) and Fable 5.1 (164.7), with overlapping confidence intervals · our sweep fired at 15:45 UTC · announced Sep 30 for Fairwind trusted testers first · the newer Gemini 4 checkpoints Business Insider reported Oct 9 file under their own hunts: Carbon (announced Oct 28, gemini-4-carbon) and Barium (gemini-4-next) · Cognition FrontierCode 2.0 (read Oct 30): 26.3% at high in Devin CLI, twelfth of 26, $6.51 a task.
Google doubles seven Gemini API prices: Gemini 4 Argon reaches Claude Opus 5.5's $4 / $20 on December 1, and Gemini 3.8 Flash goes to $1.50 / $7.50 on January 1
2026-11-06 · Google's own pricing pages date a doubling for seven Gemini models, starting with Gemini 4 Argon, which goes from $2 / $10 to $4 / $20 per million input and output tokens on December 1, the same list price as Claude Opus 5.5. Opus 5.5 is first on ArtificialWatch's bench index; Argon is seventh.
Gemini 4 Argon scores 97.5% on ARC-AGI-3 through a provider adapter, 49.8% on the standard harness
2026-10-15 · ARC Prize has published verified scores for Gemini 4 Argon, and on ARC-AGI-3 there are two very different numbers. Run through what ARC Prize calls the provider adapter harness, Argon scores 97.5% at low effort. On ARC Prize's standard harness it scores 49.8%.
Muse Wonder vs Gemini 4 Argon: quick answers
Which is better, Muse Wonder or Gemini 4 Argon?
It depends on what you measure. Of the 6 public leaderboards that rate both, Muse Wonder scores higher on 1 and Gemini 4 Argon on 5. Muse Wonder's widest lead is on LMArena; Gemini 4 Argon's widest lead is on LLM Stats. On the ArtificialWatch Index, which takes the median of each model's percentile across the boards that score it, Muse Wonder is at 68.7 and Gemini 4 Argon at 78.4 (out of 100).
Which is newer, Muse Wonder or Gemini 4 Argon?
Gemini 4 Argon. It went live on October 13, 2026, 1 day after Muse Wonder (October 12, 2026).
How do Muse Wonder and Gemini 4 Argon compare for coding?
On the ArtificialWatch coding index, Muse Wonder is at 70 (#14 of 48) and Gemini 4 Argon at 95.3 (#2). Gemini 4 Argon scores higher on all 3 coding boards that rate both.
Who makes Muse Wonder and Gemini 4 Argon?
Muse Wonder is from Meta; Gemini 4 Argon is from Google.
Where do these numbers come from?
Each score is the leaderboard's own published figure, read by ArtificialWatch (table composed November 13, 2026). Boards measure different things, so raw scores are only comparable within a row.
More comparisons
Hear when the next one drops. Free.
Free: a Chrome push and an email about 15 minutes after a new model answers on a public API. Members get a text or a phone call the moment a launch is confirmed.
- Free forever
- No card
- Unsubscribe in one click
Also: the bench index · API pricing · the wire