Llama 4 Maverick
ModelOpen sourceActiveby Meta · family “llama-maverick”
Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward... (description from the OpenRouter listing)
in: textin: imageout: textTool useStructured output
Epoch Capabilities Index
132.2 #146 of 268
90% CI 127.5 – 133.9
Released
6 Apr 2025
Context window
1.05M tokens
Max output
16K tokens
Input $ / 1M tokens
$0.19
Output $ / 1M tokens
$0.65
Cached input $ / 1M
Unknown
Each benchmark shown separately with its own source
Benchmark results (11)
| Benchmark | Domain | Score | vs best recorded | Setting | Run | Source |
|---|---|---|---|---|---|---|
| LMCA | agents | 15.9% | 24% | — | — | External ↗ |
| DTBench | reasoning | 61.9% | 63% | — | — | External ↗ |
| ARC-AGI-2 | reasoning | 0.0% | 0% | — | — | External ↗ |
| GeoBench | multimodal | 52.0% | 59% | — | — | External ↗ |
| Fiction.LiveBench | long-context | 46.2% | 48% | — | — | External ↗ |
| Lech Mazur Writing | other | 62.0% | 72% | — | — | External ↗ |
| HLE | knowledge | 5.7% | 10% | — | — | External ↗ |
| WeirdML | coding | 24.5% ±0.0 | 26% | — | — | External ↗ |
| Aider polyglot | coding | 15.6% | 18% | — | — | External ↗ |
| SimpleBench | reasoning | 27.7% | 34% | — | — | External ↗ |
| ARC-AGI | reasoning | 4.4% | 4% | — | — | External ↗ |
Source: Epoch AI Benchmarking Hub (CC BY 4.0). “External” rows are leaderboard results Epoch collects from third parties. Best reported setting per benchmark is shown.
API list price over time
Price history
$ per 1M tokens
13 observations (OpenRouter listing + Internet Archive snapshots). History accumulates with every ingest run; a single point means no change has been observed yet.
Events
No events recorded yet.