Claude Sonnet 4.5 (September 2025)
74.6%
no interval published
aiqbench · LLM benchmarks
Benchmarks
- Sources
- 8
- Entries tracked
- 22,318
- Last sync
- 22 min22 minutes ago
All sources syncing normally
- Arena · Text· 3 hours ago
- Arena · WebDev· 3 hours ago
- Arena · Vision· 3 hours ago
- Arena · Agent· 3 hours ago
- GPQA Diamond· 22 minutes ago
- SWE-bench· 22 minutes ago
- Terminal-Bench· 22 minutes ago
- ARC-AGI-2· 22 minutes ago
Filter by org:
Released:
Claude Opus 4.1 High (August 2025)
68.5%
no interval published
Claude Opus 4 High (May 2025)
64.8%
no interval published
One grid, every column. Pick axes in the Axes panel — one axis ranks like a single board, several blend into a weighted composite, all of them read as the index. Rank is assigned before any filter and never renumbers.
| Model | Org | ± CI | Swing | Add to compare | |||||
|---|---|---|---|---|---|---|---|---|---|
| 1 | Claude Sonnet 4.5 (September 2025) | anthropic | Sep 2025 | $3 / $15 | 74.6% | — | #66–232 | ||
| 2 | Claude Opus 4.1 High (August 2025) | anthropic | — | — | 68.5% | — | — | ||
| 3 | Claude Opus 4 High (May 2025) | anthropic | — | — | 64.8% | — | — | ||
| 4 | Claude-3.7 Sonnet High (February 2025) | anthropic | — | — | 64.2% | — | — | ||
| 5 | GPT-5 Medium (August 2025) | openai | — | — | 62.8% | — | — | ||
| 6 | o4-mini Low (April 2025) | openai | — | — | 58.2% | — | — | ||
| 7 | Claude Haiku 4.5 (October 2025) | anthropic | Oct 2025 | $1 / $5 | 56.4% | — | #117–611 | ||
| 8 | GPT-4.1 (April 2025) | openai | Apr 2025 | $2 / $8 | 50.3% | — | #148–546 | ||
| 9 | Gemini 2.0 Flash (February 2025) | — | — | 32.7% | — | — | |||
| 10 | o3 Medium (April 2025) | openai | — | — | 32.7% | — | — | ||
| 11 | DeepSeek R1 (January 2025) | deepseek | Jan 2025 | $0.70 / $2.50 | 30.3% | — | #117–389 | ||
| 12 | DeepSeek V3 (March 2025) | deepseek | Dec 2024 | — | 29.4% | — | #178–644 |
Showing GAIA (HAL) exactly as the source published it — its own score and confidence interval, no blending. Select a second axis to build a weighted composite.