Compare / head-to-head
Claude Opus 4.6vsClaude Opus 4.8
Claude Opus 4.8 leads 23 of 29 shared benchmarks. Both list the same combined token price. Both offer a 1M-token context window.
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
6 – 23
Claude Opus 4.8 leads
Cheaper per token
Tie
list price, input + output
Larger context
Tie
both 1M tokens
Providers
Anthropic
same provider
Claude Opus 4.6
Anthropic · released 2026-02-05
textvision
Claude Opus 4.8
Anthropic · released 2026-05-28
textvision
Quality
Benchmark matrix
29 shared · 9 only Claude Opus 4.6 · 11 only Claude Opus 4.8| Benchmark | Claude Opus 4.6 | Claude Opus 4.8 | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 29 | ||||
| SWE-bench Verified | 80.8% ↗ | 88.6% ↗ | -7.8 pt | Claude Opus 4.8 |
| GPQA Diamond | 91.3% ↗ | 93.6% ↗ | -2.3 pt | Claude Opus 4.8 |
| AIME 2026 | 96.7% ↗matharena | 100% ↗matharena ⚠ | -3.3 pt | Claude Opus 4.8 |
| LMArena Elo | 1497.5 ↗ | 1452.7 ↗ | +44.8 | Claude Opus 4.6 |
| APEX-Agents (Mercor) | 46.3% ↗ | 48.9% ↗ | -2.6 pt | Claude Opus 4.8 |
| Chess Puzzles (Epoch AI run) | 17% ↗epoch run | 34% ↗epoch run | -17 pt | Claude Opus 4.8 |
| EBR-bench (Epoch AI run) | 12.7% ↗epoch run | 28.6% ↗epoch run | -15.9 pt | Claude Opus 4.8 |
| FrontierMath Tier 4 v2 (Epoch AI run) | 26.8% ↗epoch run | 56.1% ↗epoch run | -29.3 pt | Claude Opus 4.8 |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | 66% ↗epoch run | 80% ↗epoch run | -14 pt | Claude Opus 4.8 |
| Furniture Assembly (Epoch AI run) | 28.3% ↗epoch run | 42.5% ↗epoch run | -14.2 pt | Claude Opus 4.8 |
| GSO Opt@1 (GSO) | 37.3% ↗ | 47.1% ↗ | -9.8 pt | Claude Opus 4.8 |
| LiveBench Agentic Coding (LiveBench) | 49% ↗ | 50.5% ↗ | -1.5 pt | Claude Opus 4.8 |
| LiveBench Coding (LiveBench) | 78.2% ↗ | 81.8% ↗ | -3.6 pt | Claude Opus 4.8 |
| LiveBench Data Analysis (LiveBench) | 69.9% ↗ | 66% ↗ | +3.9 pt | Claude Opus 4.6 |
| LiveBench Instruction Following (LiveBench) | 63.3% ↗ | 72% ↗ | -8.7 pt | Claude Opus 4.8 |
| LiveBench Language (LiveBench) | 83.3% ↗ | 79.7% ↗ | +3.6 pt | Claude Opus 4.6 |
| LiveBench Mathematics (LiveBench) | 89.3% ↗ | 94.3% ↗ | -5 pt | Claude Opus 4.8 |
| LiveBench Reasoning (LiveBench) | 88.7% ↗ | 89.2% ↗ | -0.5 pt | Claude Opus 4.8 |
| LMArena Vision (LMArena) | 1299.4 ↗ | 1286.4 ↗ | +13 | Claude Opus 4.6 |
| LMArena WebDev (LMArena) | 1545.9 ↗ | 1555.5 ↗ | -9.6 | Claude Opus 4.8 |
| Mystery Game Puzzles (Epoch AI run) | 25% ↗epoch run | 36% ↗epoch run | -11 pt | Claude Opus 4.8 |
| OSWorld-Verified | 72.7% ↗ | 83.4% ↗ | -10.7 pt | Claude Opus 4.8 |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 94.4% ↗epoch run | 98.3% ↗epoch run | -3.9 pt | Claude Opus 4.8 |
| SAGE (Vals AI) | 51.6% ↗ | 54.8% ↗ | -3.2 pt | Claude Opus 4.8 |
| SimpleBench (SimpleBench) | 67.6% ↗ | 64.8% ↗ | +2.8 pt | Claude Opus 4.6 |
| SimpleQA Verified | 47% ↗epoch run | 53% ↗epoch run | -6 pt | Claude Opus 4.8 |
| tau2-bench Banking Knowledge (Sierra) | 27.3% ↗ | 39.7% ↗ | -12.4 pt | Claude Opus 4.8 |
| Vending-Bench 2 (Andon Labs) | 8017.59 ↗ | 5787.43 ↗ | +2230.2 | Claude Opus 4.6 |
| WeirdML (Håvard Tveit Ihle) | 78% ↗ | 82.9% ↗ | -4.9 pt | Claude Opus 4.8 |
| Only Claude Opus 4.6 reports · 9 | ||||
| ARC-AGI-2 (Verified) | 68.8% ↗ | not reported | — | — |
| MCP Atlas | 59.5% ↗ | not reported | — | — |
| MMMLU | 91.1% ↗ | not reported | — | — |
| MMMU-Pro (no tools) | 73.9% ↗ | not reported | — | — |
| MMMU-Pro (with tools) | 77.3% ↗ | not reported | — | — |
| SWE-bench Verified (Epoch AI run) | 78.7% ↗epoch run | not reported | — | — |
| Tau2-bench Retail | 91.9% ↗ | not reported | — | — |
| Tau2-bench Telecom | 99.3% ↗ | not reported | — | — |
| Terminal-Bench 2.0 | 65.4% ↗ | not reported | — | — |
| Only Claude Opus 4.8 reports · 11 | ||||
| Humanity's Last Exam (no tools) | not reported | 49.8% ↗ | — | — |
| BrowseComp | not reported | 84.3% ↗ | — | — |
| DeepSWE v1.1 (Datacurve) | not reported | 59% ↗ | — | — |
| Humanity's Last Exam (with tools) | not reported | 57.9% ↗ | — | — |
| LMArena Agent (LMArena) | not reported | 0.0664 ↗ | — | — |
| SWE-bench Multilingual | not reported | 84.4% ↗ | — | — |
| SWE-bench Multimodal | not reported | 38.4% ↗ | — | — |
| SWE-Bench Pro | not reported | 69.2% ↗ | — | — |
| Terminal-Bench 2.1 | not reported | 74.6% ↗ | — | — |
| Terminal-Bench 4.0 (Vals AI) | not reported | 23.2% ↗ | — | — |
| Toolathlon-Verified (HKUST) | not reported | 76.2% ↗ | — | — |
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 4.6 minus Claude Opus 4.8 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 4.6 | Claude Opus 4.8 | Edge |
|---|---|---|---|
| Context window | 1M tokens | 1M tokens | Tie |
| Max output | 128K tokens | 128K tokens | Tie |
| Input price / 1M | $5 ↗ | $5 ↗ | Tie |
| Output price / 1M | $25 ↗ | $25 ↗ | Tie |
| Cached input / 1M | $0.5 ↗ | $0.5 ↗ | Tie |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $30 | $30 | Tie |
| Modalities | text · vision | text · vision | Tie |
| Released | 2026-02-05 | 2026-05-28 | — |
| Cited benchmark scores | 46 | 47 | — |
Reliability
Anthropic status
Live from /statusMore matchups
Claude Opus 4.6 vs …
Models sharing the most benchmarksMore matchups
Claude Opus 4.8 vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run Claude Opus 4.6 and Claude Opus 4.8 on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.