Claude Opus 4.5 vs Claude Opus 4.6
| Spec | Claude Opus 4.5 | Claude Opus 4.6 |
|---|---|---|
| Respan Index | 39.7 (#69) | 44.9 (#57) |
| Blended price per 1M | $10 | $10 |
| Context window | 200K | 1M |
| Input / output per 1M | $5 / $25 | $5 / $25 |
| Max output | 64K | 128K |
| Input | text, vision | text, vision |
| Released | 2025-11-24 | 2026-02-05 |
Respan Index by category
Each category is scored 0 to 100, and how far ahead the winner is. Leads under a point are close to even.
| Category | Claude Opus 4.5 | Claude Opus 4.6 |
|---|---|---|
| Coding | 19.8 Claude Opus 4.5 | 29.0 Claude Opus 4.6 |
| Agents | 53.7 Claude Opus 4.5 | 55.7 Claude Opus 4.6 |
| Reasoning | 35.0 Claude Opus 4.5 | 42.7 Claude Opus 4.6 |
| Knowledge & language | 45.4 Claude Opus 4.5 | 47.4 Claude Opus 4.6 |
| Human preference | 53.5 Claude Opus 4.5 | 55.1 Claude Opus 4.6 |
Benchmarks both report
Claude Opus 4.6 leads on 22 of the 30 benchmarks both report. Higher is better. Each score links to where it was published.
| Benchmark | Claude Opus 4.5 | Claude Opus 4.6 |
|---|---|---|
| LMArena Elo | 1474 | 1503 |
| LMArena WebDev (LMArena) | 1494 | 1546 |
| Chess Puzzles (Epoch AI) | 14% | 21% |
| EBR-bench (Epoch AI) | 14.3% | 12.7% |
| FrontierMath Tier 4 v2 (Epoch AI) | 4.9% | 26.8% |
| FrontierMath Tiers 1-3 v2 (Epoch AI) | 34.4% | 66% |
| Furniture Assembly (Epoch AI) | 28.3% | 28.3% |
| GPQA Diamond | 87% | 91.3% |
| LiveBench Agentic Coding (LiveBench) | 39.7% | 49% |
| LiveBench Coding (LiveBench) | 79.7% | 78.2% |
| LiveBench Data Analysis (LiveBench) | 74.4% | 69.9% |
| LiveBench Instruction Following (LiveBench) | 62.5% | 63.3% |
| LiveBench Language (LiveBench) | 81.3% | 83.3% |
| LiveBench Mathematics (LiveBench) | 90.4% | 89.3% |
| LiveBench Reasoning (LiveBench) | 80.1% | 88.7% |
| Mystery Game Puzzles (Epoch AI) | 22% | 25% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 86.1% | 94.4% |
| SAGE (Vals AI) | 52.1% | 51.6% |
| SimpleQA Verified | 45.7% | 47% |
| SWE-bench Verified (Epoch AI) | 76.7% | 78.7% |
| Vending-Bench 2 (Andon Labs) | 4967.06 | 8017.59 |
| WeirdML (Håvard Tveit Ihle) | 63.7% | 78% |
| tau2-bench Banking Knowledge (Sierra) | 24.7% | 27.3% |
| GSO Opt@1 (GSO) | 24.5% | 37.3% |
| ARC-AGI-2 (Verified) | 37.6% | 68.8% |
| MCP Atlas | 62.3% | 59.5% |
| MMMLU | 90.8% | 91.1% |
| SWE-bench Verified | 80.9% | 80.8% |
| Terminal-Bench 2.0 | 59.3% | 65.4% |
| SimpleBench (SimpleBench) | 62% | 67.6% |
FAQ
Is Claude Opus 4.5 better than Claude Opus 4.6?
Claude Opus 4.6 scores 44.9 on the Respan Index (#57), ahead of Claude Opus 4.5 at 39.7 (#69). Claude Opus 4.6 leads on 22 of the 30 benchmarks both report.
Which is cheaper, Claude Opus 4.5 or Claude Opus 4.6?
Claude Opus 4.5 lists at $5 input and $25 output per 1M tokens; Claude Opus 4.6 at $5 and $25. Both cost about the same per token at list price.
Which has the larger context window?
Claude Opus 4.5 accepts 200K tokens and Claude Opus 4.6 1M tokens.
Compare other models
- vsClaude Opus 4.5 vs Claude Opus 4.1
- vsClaude Opus 4.7 vs Claude Opus 4.6
- vsClaude Sonnet 5.5 vs Claude Opus 4.5
- vsClaude Opus 4.5 vs GPT-5
- vsClaude Opus 4.5 vs Dola Seed 2.0 Pro
- vsClaude Opus 4.5 vs Gemini 3 Pro Preview
- vsClaude Opus 4.5 vs Qwen3.6-Max-Preview
- vsClaude Opus 4.6 vs DeepSeek V4 Flash
- vsClaude Opus 4.6 vs Qwen3.8-Flash-Next
- vsClaude Opus 4.6 vs GLM-5.3-Flash