New: Claude Fable 5.2 eval report is here.Read the report ↗

Best LLMs for coding in Cursor

CursorBench

Tests whether an agent can complete ambiguous, multi-file coding tasks from real Cursor sessions, helping you choose a model for client development work with less step-by-step direction. Scores, effort levels and task costs are specific to Cursor's agent setup.

Updated: Data as of:
Widely usedNot saturated
Official benchmark
CursorBench results. Each row identifies its benchmark version and evaluation track. Scores from different versions are not directly comparable. Best effort from the latest stored report date per model and version.
Model / creatorVersionReported score
Claude Fable 5.2Anthropic · Proprietary4.0 / Cursor agent60%
Claude Opus 5.5Anthropic · Proprietary4.0 / Cursor agent57.8%
Claude Sonnet 5.5Anthropic · Proprietary4.0 / Cursor agent55.5%
Gemini 4 ArgonGoogle · Proprietary4.0 / Cursor agent52.9%
Claude Fable 5.1Anthropic · Proprietary4.0 / Cursor agent51.8%
Claude Haiku 5.5Anthropic · Proprietary4.0 / Cursor agent48.4%
DeepSeek V4.1 Flash 1017DeepSeek · Open source / open weights4.0 / Cursor agent47.7%
Claude Opus 5Anthropic · Proprietary4.0 / Cursor agent46.6%
Grok 4.7xAI · Proprietary4.0 / Cursor agent46.3%
GLM-5.3Z.ai · Open source / open weights4.0 / Cursor agent42.6%
GPT-5.6 SolOpenAI · Proprietary4.0 / Cursor agent41.7%
Muse Spark 1.3Meta · Proprietary4.0 / Cursor agent41.6%
GPT-5.6 TerraOpenAI · Proprietary4.0 / Cursor agent41.3%
Gemini 3.8 FlashGoogle · Proprietary4.0 / Cursor agent39.6%
GLM-5.3-FlashZ.ai · Open source / open weights4.0 / Cursor agent36.8%
GPT-5.6 LunaOpenAI · Proprietary4.0 / Cursor agent35.9%
Claude Sonnet 5Anthropic · Proprietary4.0 / Cursor agent34.1%
View all 75 published reports
4.0 / Cursor agent

cost usd per task: 32.79 · effort: max · harness: Cursor production agent · output tokens per task: 352794 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 218 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 13.43 · effort: max · harness: Cursor production agent · output tokens per task: 218363 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 185 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 9.7 · effort: high · harness: Cursor production agent · output tokens per task: 99259 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 69 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 15.48 · effort: xhigh · harness: Cursor production agent · output tokens per task: 161251 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 108 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 3.97 · effort: high · harness: Cursor production agent · output tokens per task: 53078 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 68 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 6.98 · effort: xhigh · harness: Cursor production agent · output tokens per task: 101083 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 109 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 7.05 · effort: max · harness: Cursor production agent · output tokens per task: 271920 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 170 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 4.66 · effort: medium · harness: Cursor production agent · output tokens per task: 44103 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 35 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 3.94 · effort: low · harness: Cursor production agent · output tokens per task: 36616 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 30 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 2.81 · effort: xhigh · harness: Cursor production agent · output tokens per task: 100158 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 78 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 4.57 · effort: high · harness: Cursor production agent · output tokens per task: 109500 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 118 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 4.39 · effort: medium · harness: Cursor production agent · output tokens per task: 103600 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 115 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 2.91 · effort: medium · harness: Cursor production agent · output tokens per task: 37954 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 54 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 17.28 · effort: max · harness: Cursor production agent · output tokens per task: 117236 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 128 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 13.01 · effort: xhigh · harness: Cursor production agent · output tokens per task: 87294 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 101 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 4.09 · effort: low · harness: Cursor production agent · output tokens per task: 94200 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 110 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 9.08 · effort: high · harness: Cursor production agent · output tokens per task: 58438 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 77 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 1.12 · effort: max · harness: Cursor production agent · output tokens per task: 325934 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 163 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 1.2 · effort: high · harness: Cursor production agent · output tokens per task: 37391 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 41 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.98 · effort: max · harness: Cursor production agent · output tokens per task: 537047 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 187 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 7.05 · effort: medium · harness: Cursor production agent · output tokens per task: 45411 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 63 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 11.95 · effort: max · harness: Cursor production agent · output tokens per task: 85384 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 106 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.8 · effort: medium · harness: Cursor production agent · output tokens per task: 398255 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 189 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 6.01 · effort: xhigh · harness: Cursor production agent · output tokens per task: 70141 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 88 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 11.43 · effort: xhigh · harness: Cursor production agent · output tokens per task: 80094 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 103 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.77 · effort: high · harness: Cursor production agent · output tokens per task: 385196 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 186 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 5.44 · effort: low · harness: Cursor production agent · output tokens per task: 34795 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 51 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.6 · effort: low · harness: Cursor production agent · output tokens per task: 290651 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 161 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 9 · effort: high · harness: Cursor production agent · output tokens per task: 61405 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 86 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.56 · effort: xhigh · harness: Cursor production agent · output tokens per task: 143813 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 93 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 4.69 · effort: high · harness: Cursor production agent · output tokens per task: 56382 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 71 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 1.17 · effort: low · harness: Cursor production agent · output tokens per task: 15811 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 28 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 6.94 · effort: medium · harness: Cursor production agent · output tokens per task: 45272 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 72 · Tools enabled

Cursor ↗
GLM-5.342.6%
4.0 / Cursor agent

cost usd per task: 5.05 · effort: max · harness: Cursor production agent · output tokens per task: 96387 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 166 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.32 · effort: high · harness: Cursor production agent · output tokens per task: 77057 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 59 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 8.23 · effort: max · harness: Cursor production agent · output tokens per task: 42944 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 99 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 2.64 · effort: max · harness: Cursor production agent · output tokens per task: 52005 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 98 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 3.49 · effort: medium · harness: Cursor production agent · output tokens per task: 36683 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 60 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 5.14 · effort: max · harness: Cursor production agent · output tokens per task: 60814 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 107 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 4.87 · effort: low · harness: Cursor production agent · output tokens per task: 31995 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 57 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 4.7 · effort: high · harness: Cursor production agent · output tokens per task: 162565 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 324 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.52 · effort: medium · harness: Cursor production agent · output tokens per task: 16036 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 22 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 3.24 · effort: high · harness: Cursor production agent · output tokens per task: 60031 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 114 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 4.4 · effort: xhigh · harness: Cursor production agent · output tokens per task: 24729 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 55 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 2.1 · effort: xhigh · harness: Cursor production agent · output tokens per task: 40891 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 83 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 4.06 · effort: medium · harness: Cursor production agent · output tokens per task: 128364 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 290 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.17 · effort: medium · harness: Cursor production agent · output tokens per task: 42659 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 39 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.39 · effort: max · harness: Cursor production agent · output tokens per task: 56410 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 118 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 1.03 · effort: max · harness: Cursor production agent · output tokens per task: 87284 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 208 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.37 · effort: low · harness: Cursor production agent · output tokens per task: 11668 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 18 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 2.85 · effort: high · harness: Cursor production agent · output tokens per task: 16174 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 41 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 7.17 · effort: max · harness: Cursor production agent · output tokens per task: 149257 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 140 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 1.81 · effort: xhigh · harness: Cursor production agent · output tokens per task: 23436 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 43 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 1.66 · effort: high · harness: Cursor production agent · output tokens per task: 30654 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 69 · Tools enabled

Cursor ↗
GLM-5.333.3%
4.0 / Cursor agent

cost usd per task: 2.04 · effort: low · harness: Cursor production agent · output tokens per task: 31983 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 81 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 1.58 · effort: low · harness: Cursor production agent · output tokens per task: 15677 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 40 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.44 · effort: xhigh · harness: Cursor production agent · output tokens per task: 40598 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 98 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 1.49 · effort: medium · harness: Cursor production agent · output tokens per task: 27255 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 64 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 4.55 · effort: xhigh · harness: Cursor production agent · output tokens per task: 83373 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 102 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.25 · effort: high · harness: Cursor production agent · output tokens per task: 35104 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 84 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 1.77 · effort: medium · harness: Cursor production agent · output tokens per task: 10111 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 32 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.08 · effort: low · harness: Cursor production agent · output tokens per task: 23382 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 25 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 3.48 · effort: high · harness: Cursor production agent · output tokens per task: 61146 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 85 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 1.11 · effort: high · harness: Cursor production agent · output tokens per task: 13162 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 33 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.25 · effort: high · harness: Cursor production agent · output tokens per task: 23368 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 64 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.93 · effort: low · harness: Cursor production agent · output tokens per task: 17483 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 47 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 2.31 · effort: medium · harness: Cursor production agent · output tokens per task: 39114 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 65 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.64 · effort: medium · harness: Cursor production agent · output tokens per task: 7307 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 25 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.15 · effort: low · harness: Cursor production agent · output tokens per task: 17831 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 58 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.52 · effort: low · harness: Cursor production agent · output tokens per task: 5914 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 23 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.87 · effort: low · harness: Cursor production agent · output tokens per task: 4885 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 21 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.56 · effort: minimal · harness: Cursor production agent · output tokens per task: 10620 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 34 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 1.39 · effort: low · harness: Cursor production agent · output tokens per task: 23772 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 46 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.08 · effort: medium · harness: Cursor production agent · output tokens per task: 7642 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 32 · Tools enabled

Cursor ↗
4.0 / Cursor agent

cost usd per task: 0.03 · effort: low · harness: Cursor production agent · output tokens per task: 3288 · Publisher table observed November 9; row publication date unspecified; current harness may differ from earlier system cards; small score differences may not be significant. · steps per task: 18 · Tools enabled

Cursor ↗