Claude Sonnet 5 vs Grok 4.5
| Spec | Claude Sonnet 5 | Grok 4.5 |
|---|---|---|
| Respan Index | 46.7 (#39) | 46.6 (#40) |
| Blended price per 1M | $4 | $3 |
| Context window | 1M | 500K |
| Input / output per 1M | $2 / $10 | $2 / $6 |
| Max output | 128K | - |
| Input | text, vision | text, vision |
| Released | 2026-06-30 | 2026-07-08 |
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 Sonnet 5 | Grok 4.5 |
|---|---|---|
| Coding | 26.9 Claude Sonnet 5 | 26.6 Grok 4.5 |
| Agents | 62.8 Claude Sonnet 5 | 63.9 Grok 4.5 |
| Reasoning | 51.0 Claude Sonnet 5 | 43.3 Grok 4.5 |
| Knowledge & language | 34.0 Claude Sonnet 5 | 49.7 Grok 4.5 |
| Human preference | 54.4 Claude Sonnet 5 | 54.8 Grok 4.5 |
Benchmarks both report
They split the 25 benchmarks both report evenly. Higher is better. Each score links to where it was published.
| Benchmark | Claude Sonnet 5 | Grok 4.5 |
|---|---|---|
| Chess Puzzles (Epoch AI) | 45% | 35% |
| LMArena Agent (LMArena) | 0.0294 | 0.0079 |
| LMArena Elo | 1462 | 1465 |
| LMArena WebDev (LMArena) | 1540 | 1553 |
| LMArena Vision (LMArena) | 1262 | 1279 |
| Terminal-Bench 4.0 (Vals AI) | 10.6% | 8.6% |
| APEX-Agents (Mercor) | 54.5% | 56.2% |
| DeepSWE v1.1 (Datacurve) | 53.8% | 53.8% |
| FrontierMath Tier 4 v2 (Epoch AI) | 29.3% | 24.4% |
| FrontierMath Tiers 1-3 v2 (Epoch AI) | 65.6% | 57.2% |
| GPQA Diamond | 90.5% | 93.4% |
| LiveBench Agentic Coding (LiveBench) | 59.4% | 56.5% |
| LiveBench Coding (LiveBench) | 80.7% | 68.6% |
| LiveBench Data Analysis (LiveBench) | 71.7% | 73% |
| LiveBench Instruction Following (LiveBench) | 63.9% | 71.5% |
| LiveBench Language (LiveBench) | 75% | 82.8% |
| LiveBench Mathematics (LiveBench) | 92.9% | 90.8% |
| LiveBench Reasoning (LiveBench) | 88.7% | 87.2% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 94.7% | 97.8% |
| SAGE (Vals AI) | 48.9% | 35% |
| SimpleQA Verified | 33.7% | 48.3% |
| SWE-rebench 2026-05-15 to 2026-07-01 (Nebius) | 56.8% | 63.8% |
| Vending-Bench 2 (Andon Labs) | 6377.7 | 3887.43 |
| WeirdML (Håvard Tveit Ihle) | 68.8% | 46.4% |
| SimpleBench (SimpleBench) | 60.6% | 70% |
FAQ
Is Claude Sonnet 5 better than Grok 4.5?
Claude Sonnet 5 scores 46.7 on the Respan Index (#39), ahead of Grok 4.5 at 46.6 (#40). They split the 25 benchmarks both report evenly.
Which is cheaper, Claude Sonnet 5 or Grok 4.5?
Claude Sonnet 5 lists at $2 input and $10 output per 1M tokens; Grok 4.5 at $2 and $6. Grok 4.5 is 1.3x cheaper per token at list price.
Which has the larger context window?
Claude Sonnet 5 accepts 1M tokens and Grok 4.5 500K tokens.