CursorBench
Cursor evaluates agents on ambiguous, multi-file tasks taken from real Cursor sessions.
- Unit
- score
- Results
- 78 results18 Tokscale models matched, 18 in the source
- Last update
- Oct 12, 2026Newest model release
- Retrieved
- Oct 21, 2026Source data through Oct 15, 2026
Conditions. Run by Cursor at each reasoning level. Cost, tokens and steps are averages per task.
Attribution. CursorBench results by Cursor, collected by Epoch AI, ‘Capabilities & benchmarking’, epoch.ai/benchmarks (CC BY 4.0)
Results
One row per model: its best reported result. The rank is the position in this list, by score; the source's own ranking can differ.
Tokscale usage counts accounts that used the model in public Tokscale submissions, Oct 14 – Nov 12, 2026. It is shown next to the score and does not explain it.
| Rank | Model | Score (score) | Conditions | Cost and tokens | Date | Tokscale usage (30 days) | Source link |
|---|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5Listed as Opus 5.5 Max | 57.8% | Max effort | $13.43 per task218K tokens per task | Sep 22, 2026Model released | 709active users | Source for Claude Opus 5.5 (opens in a new tab) |
| 2 | Claude Sonnet 5.5Listed as Sonnet 5.5 Max | 55.5% | Max effort | $7.05 per task272K tokens per task | Sep 28, 2026Model released | 555active users | Source for Claude Sonnet 5.5 (opens in a new tab) |
| 3 | Gemini 4 ArgonListed as Gemini 4 Argon High | 52.9% | High effort | $4.57 per task110K tokens per task | Sep 30, 2026Model released | 82active users | Source for Gemini 4 Argon (opens in a new tab) |
| 4 | Claude Fable 5.1Listed as Fable 5.1 Max | 51.8% | Max effort | $17.28 per task117K tokens per task | Sep 1, 2026Model released | 361active users | Source for Claude Fable 5.1 (opens in a new tab) |
| 5 | Muse Wonder 1.0Listed as Muse Wonder Max | 50.4% | Max effort | $5.25 per task49.7K tokens per task | Oct 12, 2026Model released | 17active users | Source for Muse Wonder 1.0 (opens in a new tab) |
| 6 | Claude Haiku 5.5Listed as Haiku 5.5 Max | 48.4% | Max effort | $1.12 per task326K tokens per task | Oct 7, 2026Model released | 477active users | Source for Claude Haiku 5.5 (opens in a new tab) |
| 7 | Claude Opus 5Listed as Opus 5 Max | 46.6% | Max effort | $11.95 per task85.4K tokens per task | Jul 24, 2026Model released | 163active users | Source for Claude Opus 5 (opens in a new tab) |
| 8 | Grok 4.7Listed as Grok 4.7 Extra High | 46.3% | XHigh effort | $6.01 per task70.1K tokens per task | Sep 21, 2026Model released | 131active users | Source for Grok 4.7 (opens in a new tab) |
| 9 | GLM-5.3Listed as GLM 5.3 Max | 42.6% | Max effort | $5.05 per task96.4K tokens per task | Aug 14, 2026Model released | 138active users | Source for GLM-5.3 (opens in a new tab) |
| 10 | GPT-5.6 SolListed as GPT-5.6 Sol Max | 41.7% | Max effort | $8.23 per task42.9K tokens per task | Jul 9, 2026Model released | 217active users | Source for GPT-5.6 Sol (opens in a new tab) |
| 11 | Muse Spark 1.3Listed as Muse Spark 1.3 Max | 41.6% | Max effort | $2.64 per task52K tokens per task | Sep 2, 2026Model released | 21active users | Source for Muse Spark 1.3 (opens in a new tab) |
| 12 | Grok 4.6Listed as Grok 4.6 Extra High | 41.4% | XHigh effort | $6.10 per task49.8K tokens per task | Aug 12, 2026Model released | 41active users | Source for Grok 4.6 (opens in a new tab) |
| 13 | GPT-5.6 TerraListed as GPT-5.6 Terra Max | 41.3% | Max effort | $5.14 per task60.8K tokens per task | Jul 9, 2026Model released | 119active users | Source for GPT-5.6 Terra (opens in a new tab) |
| 14 | Gemini 3.8 FlashListed as Gemini 3.8 Flash High | 39.6% | High effort | $4.70 per task163K tokens per task | Sep 2, 2026Model released | 66active users | Source for Gemini 3.8 Flash (opens in a new tab) |
| 15 | GLM-5.3 FlashListed as GLM 5.3 Flash Max | 36.8% | Max effort | $0.39 per task56.4K tokens per task | Aug 20, 2026Model released | 170active users | Source for GLM-5.3 Flash (opens in a new tab) |
| 16 | GPT-5.6 LunaListed as GPT-5.6 Luna Max | 35.9% | Max effort | $1.03 per task87.3K tokens per task | Jul 9, 2026Model released | 158active users | Source for GPT-5.6 Luna (opens in a new tab) |
| 17 | Claude Sonnet 5Listed as Sonnet 5 Max | 34.1% | Max effort | $7.17 per task149K tokens per task | Jun 30, 2026Model released | 138active users | Source for Claude Sonnet 5 (opens in a new tab) |
| 18 | 27.7% | Not reported | $0.68 per task17.3K tokens per task | May 18, 2026Model released | 9active users | Source for Composer 2.5 (opens in a new tab) |
Showing 18 of 18 models
Score by release date
Each dot is one model's best result on CursorBench, placed at the model's release date. The dashed line is the best score among models released up to that date.
- Used on Tokscale in the last 30 days
- No Tokscale usage in the last 30 days
- Best score so far
Claude Opus 5.5: 57.8% · released Sep 22, 2026 · Max effort · 709 active users on Tokscale in the last 30 days
18 models, released May 18, 2026 to Oct 12, 2026. Highest: Claude Opus 5.5, 57.8%, released Sep 22, 2026. Use the left and right arrow keys to read each model in release order.20%30%40%50%60%
Show data
| Model | Released | Score | Tokscale users, 30 days |
|---|---|---|---|
| Muse Wonder 1.0 | Oct 12, 2026 | 50.4%Max effort | 17 |
| Claude Haiku 5.5 | Oct 7, 2026 | 48.4%Max effort | 477 |
| Gemini 4 Argon | Sep 30, 2026 | 52.9%High effort | 82 |
| Claude Sonnet 5.5 | Sep 28, 2026 | 55.5%Max effort | 555 |
| Claude Opus 5.5 | Sep 22, 2026 | 57.8%Max effort | 709 |
| Grok 4.7 | Sep 21, 2026 | 46.3%XHigh effort | 131 |
| Gemini 3.8 Flash | Sep 2, 2026 | 39.6%High effort | 66 |
| Muse Spark 1.3 | Sep 2, 2026 | 41.6%Max effort | 21 |
| Claude Fable 5.1 | Sep 1, 2026 | 51.8%Max effort | 361 |
| GLM-5.3 Flash | Aug 20, 2026 | 36.8%Max effort | 170 |
| GLM-5.3 | Aug 14, 2026 | 42.6%Max effort | 138 |
| Grok 4.6 | Aug 12, 2026 | 41.4%XHigh effort | 41 |
| Claude Opus 5 | Jul 24, 2026 | 46.6%Max effort | 163 |
| GPT-5.6 Luna | Jul 9, 2026 | 35.9%Max effort | 158 |
| GPT-5.6 Terra | Jul 9, 2026 | 41.3%Max effort | 119 |
| GPT-5.6 Sol | Jul 9, 2026 | 41.7%Max effort | 217 |
| Claude Sonnet 5 | Jun 30, 2026 | 34.1%Max effort | 138 |
| Composer 2.5 | May 18, 2026 | 27.7% | 9 |
Method and matching
How results are run
- Run by Cursor at each reasoning level. Cost, tokens and steps are averages per task.
- Results are Epoch AI's copy of Cursor's published table and can lag the live page.
Matching to Tokscale models
- 78 of 78 results are matched to 18 Tokscale models: 77 by exact id, 1 by normalized name.
- No match is marked as reviewed by hand yet.
- Reasoning effort, thinking budget and agent are conditions of a result. They never change which model a result belongs to.
License and attribution
- CC BY 4.0 (opens in a new tab), retrieved Oct 21, 2026.
- CursorBench results by Cursor, collected by Epoch AI, ‘Capabilities & benchmarking’, epoch.ai/benchmarks (CC BY 4.0)
Sources and attribution
Each score keeps its own source, unit and run conditions. Tokscale does not combine benchmarks into one score, and a missing score means the model was not evaluated, not that it scored zero.