Compare / head-to-head
Claude Opus 4vsClaude Opus 5
Claude Opus 5 leads 2 of 2 shared benchmarks. Claude Opus 5 is 3x cheaper per token. Claude Opus 5 has the larger context window (1M tokens).
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
0 – 2
Claude Opus 5 leads
Cheaper per token
Claude Opus 5
3x cheaper, input + output
Larger context
Claude Opus 5
1M tokens
Provider uptime (30d)
100%
Anthropic
Claude Opus 4
Anthropic · released 2025-05-22
textvision
Claude Opus 5
Anthropic · released 2026-07-24
textvision
Quality
Benchmark matrix
2 shared · 4 only Claude Opus 4 · 7 only Claude Opus 5| Benchmark | Claude Opus 4 | Claude Opus 5 | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 2 | ||||
| GPQA Diamond | 76.3% ↗epoch | 93.9% ↗epoch | -17.6 pt | Claude Opus 5 |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 64.4% ↗epoch | 98.9% ↗epoch | -34.5 pt | Claude Opus 5 |
| Only Claude Opus 4 reports · 4 | ||||
| SWE-bench Verified | 72.5% ↗ | not reported | — | — |
| AIME 2024 (no extended thinking) | 33.9% ↗ | not reported | — | — |
| SWE-bench Verified (Epoch AI run) | 70.7% ↗epoch | not reported | — | — |
| Terminal-bench | 43.2% ↗ | not reported | — | — |
| Only Claude Opus 5 reports · 7 | ||||
| Humanity's Last Exam (no tools) | not reported | 56.6% ↗ | — | — |
| CursorBench 3.2.0 | not reported | 70% ↗ | — | — |
| FrontierMath Tier 4 v2 (Epoch AI run) | not reported | 73.2% ↗epoch | — | — |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | not reported | 85.6% ↗epoch | — | — |
| GDPval-AA v2 | not reported | 1824 ↗ | — | — |
| SimpleQA Verified | not reported | 59.9% ↗epoch | — | — |
| Terminal-Bench Science 0.1 | not reported | 29% ↗ | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is Claude Opus 4 minus Claude Opus 5 in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | Claude Opus 4 | Claude Opus 5 | Edge |
|---|---|---|---|
| Context window | 200K tokens | 1M tokens | Claude Opus 5 |
| Max output | 32K tokens | 128K tokens | Claude Opus 5 |
| Input price / 1M | $15 ↗ | $5 ↗ | Claude Opus 5 |
| Output price / 1M | $75 ↗ | $25 ↗ | Claude Opus 5 |
| Cached input / 1M | $1.5 ↗ | $0.5 ↗ | Claude Opus 5 |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $90 | $30 | Claude Opus 5 |
| Modalities | text · vision | text · vision | Tie |
| Released | 2025-05-22 | 2026-07-24 | — |
| Cited benchmark scores | 6 | 9 | — |
Reliability
Anthropic status
Last 60 days · refreshed daily on this page · live on /statusMore matchups
Claude Opus 4 vs …
Models sharing the most benchmarksMore matchups
Claude Opus 5 vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run Claude Opus 4 and Claude Opus 5 on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.