Claude Opus 5
ModelActiveby Anthropic · family “claude-opus”
Claude Opus 5 is Anthropic’s flagship model for demanding reasoning, coding, and long-horizon agentic work. It is particularly strong at end-to-end software tasks, code review and bug finding, visual analysis... (description from the OpenRouter listing)
in: textin: imagein: fileout: textReasoningTool useStructured output
Epoch Capabilities Index
162.7 #4 of 268
90% CI 160.0 – 166.5
Released
24 Jul 2026
Context window
1M tokens
Max output
128K tokens
Input $ / 1M tokens
$5.00
Output $ / 1M tokens
$25.0
Cached input $ / 1M
$0.50
Each benchmark shown separately with its own source
Benchmark results (21)
| Benchmark | Domain | Score | vs best recorded | Setting | Run | Source |
|---|---|---|---|---|---|---|
| Furniture Assembly | multimodal | 60.8% ±6.2 | 73% | max | 10 Sep 2026 | Epoch ↗ |
| FrontierSWE | other | 52.0% | 79% | max | — | External ↗ |
| LMCA | agents | 63.7% | 93% | high | — | External ↗ |
| DTBench | reasoning | 97.9% | 99% | high | — | External ↗ |
| Mystery Game Puzzles | games | 59.0% ±4.9 | 70% | max | 25 Jul 2026 | Epoch ↗ |
| EBR-bench | reasoning | 45.7% ±3.6 | 60% | max | 5 Sep 2026 | Epoch ↗ |
| OSWorld 2.0 | agents | 31.4% | 100% | max | — | External ↗ |
| FrontierMath-Tiers-1-3-v2-Private | math | 85.6% ±2.1 | 91% | max | 24 Jul 2026 | Eval log ↗ |
| FrontierMath-Tier-4-v2-Private | math | 73.2% ±7.0 | 75% | max | 24 Jul 2026 | Eval log ↗ |
| FrontierCode | coding | 53.4% | 100% | max | — | External ↗ |
| DeepSWE | coding | 73.6% | 99% | max | — | External ↗ |
| ProofBench | math | 99.0% | 99% | max | — | External ↗ |
| APEX-Agents | agents | 65.8% | 87% | max | — | External ↗ |
| Chess Puzzles | games | 42.0% ±5.0 | 58% | max | 24 Jul 2026 | Epoch ↗ |
| SimpleQA Verified | knowledge | 59.9% ±1.6 | 79% | max | 10 Aug 2026 | Epoch ↗ |
| ARC-AGI-2 | reasoning | 90.4% | 95% | max | — | External ↗ |
| WeirdML | coding | 91.8% ±0.0 | 98% | max | — | External ↗ |
| OTIS Mock AIME 2024-2025 | math | 98.9% ±1.1 | 99% | max | 24 Jul 2026 | Epoch ↗ |
| Balrog | games | 63.4% | 93% | max | — | External ↗ |
| GPQA diamond | science | 93.9% ±1.5 | 98% | max | 24 Jul 2026 | Epoch ↗ |
| ARC-AGI | reasoning | 97.5% | 99% | max | — | 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
3 recorded prices on 3 Aug 2026 (OpenRouter listing + Internet Archive snapshots); re-read on every data refresh, most recently 40m ago. Steps show when the price changed.
Events
- BenchmarkEpoch AI evaluates Claude Opus 5Epoch AI Benchmarking Hub
- BenchmarkEpoch AI evaluates Claude Opus 5Epoch AI Benchmarking Hub
- Model launchMajorAnthropic releases Claude Opus 5Epoch AI Benchmarking Hub