Skip to content
Weights Not Walls
Menu

CursorBench

Reported

No description yet: we only describe an eval set when we can do it correctly.

Kind
Published by a third party's leaderboard, collected by Epoch AI. The leaderboard is named and linked.
Link
https://cursor.com/cursorbench
Score shown
Epoch's column "Score", as a percentage, higher is better. The source gives no error bars.
Results
94 in all: 10 for open-weight models, 84 for closed ones. 10 linked to a model page here.

Best open-weight result: deepseek-ai/DeepSeek-V4.1-Flash-1017, 47.7%, number 28 of 94 results.

CursorBench results
#ModelWeightsSettingScoreDateSource
1gemini-4-carbon_highGoogle DeepMindclosedeffort high; Reasoning level: High60.4%not datedcursor.com
2gemini-4-carbon_mediumGoogle DeepMindclosedeffort medium; Reasoning level: Medium60.1%not datedcursor.com
3claude-fable-5-2_maxAnthropicclosedeffort max; Reasoning level: Max60.0%not datedcursor.com
4gemini-4-carbon_lowGoogle DeepMindclosedeffort low; Reasoning level: Low58.6%not datedcursor.com
5claude-opus-5-5_maxAnthropicclosedeffort max; Reasoning level: Max57.8%not datedcursor.com
6gemini-4-hydrogen_xhighGoogle DeepMindclosedeffort xhigh; Reasoning level: Extra High57.8%not datedcursor.com
7claude-fable-5-2_highAnthropicclosedeffort high; Reasoning level: High56.9%not datedcursor.com
8claude-fable-5-2_xhighAnthropicclosedeffort xhigh; Reasoning level: Extra High56.3%not datedcursor.com
9claude-opus-5-5_highAnthropicclosedeffort high; Reasoning level: High56.0%not datedcursor.com
10claude-opus-5-5_xhighAnthropicclosedeffort xhigh; Reasoning level: Extra High56.0%not datedcursor.com
11claude-sonnet-5-5_maxAnthropicclosedeffort max; Reasoning level: Max55.5%not datedcursor.com
12claude-fable-5-2_mediumAnthropicclosedeffort medium; Reasoning level: Medium54.8%not datedcursor.com
13gemini-4-hydrogen_highGoogle DeepMindclosedeffort high; Reasoning level: High54.3%not datedcursor.com
14claude-fable-5-2_lowAnthropicclosedeffort low; Reasoning level: Low53.2%not datedcursor.com
15claude-sonnet-5-5_xhighAnthropicclosedeffort xhigh; Reasoning level: Extra High53.1%not datedcursor.com
16gemini-4-argon_highGoogle DeepMindclosedeffort high; Reasoning level: High52.9%not datedcursor.com
17gemini-4-argon_mediumGoogle DeepMindclosedeffort medium; Reasoning level: Medium52.8%not datedcursor.com
18claude-opus-5-5_mediumAnthropicclosedeffort medium; Reasoning level: Medium52.5%not datedcursor.com
19claude-fable-5-1_maxAnthropicclosedeffort max; Reasoning level: Max51.8%not datedcursor.com
20claude-fable-5-1_xhighAnthropicclosedeffort xhigh; Reasoning level: Extra High51.6%not datedcursor.com
21gemini-4-argon_lowGoogle DeepMindclosedeffort low; Reasoning level: Low50.6%not datedcursor.com
22muse-wonder-1.0_maxMeta AIclosedeffort max; Reasoning level: Max50.4%not datedcursor.com
23claude-fable-5-1_highAnthropicclosedeffort high; Reasoning level: High49.2%not datedcursor.com
24gemini-4-argon_minimalGoogle DeepMindclosedeffort minimal48.8%not datedcursor.com
25claude-haiku-5-5_maxAnthropicclosedeffort max; Reasoning level: Max48.4%not datedcursor.com
26muse-wonder-1.0_xhighMeta AIclosedeffort xhigh; Reasoning level: Extra High48.2%not datedcursor.com
27claude-sonnet-5-5_highAnthropicclosedeffort high; Reasoning level: High47.8%not datedcursor.com
28deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_max, DeepSeekopeneffort max; Reasoning level: Max47.7%not datedcursor.com
29claude-fable-5-1_mediumAnthropicclosedeffort medium; Reasoning level: Medium46.8%not datedcursor.com
30gemini-4-hydrogen_mediumGoogle DeepMindclosedeffort medium; Reasoning level: Medium46.7%not datedcursor.com
31claude-opus-5_maxAnthropicclosedeffort max; Reasoning level: Max46.6%not datedcursor.com
32deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_medium, DeepSeekopeneffort medium; Reasoning level: Medium46.3%not datedcursor.com
33grok-4.7_xhighxAIclosedeffort xhigh; Reasoning level: Extra High46.3%not datedcursor.com
34claude-opus-5_xhighAnthropicclosedeffort xhigh; Reasoning level: Extra High46.1%not datedcursor.com
35muse-wonder-1.0_highMeta AIclosedeffort high; Reasoning level: High46.1%not datedcursor.com
36deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_high, DeepSeekopeneffort high; Reasoning level: High45.6%not datedcursor.com
37muse-wonder-1.0_mediumMeta AIclosedeffort medium; Reasoning level: Medium45.2%not datedcursor.com
38claude-fable-5-1_lowAnthropicclosedeffort low; Reasoning level: Low45.1%not datedcursor.com
39deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_low, DeepSeekopeneffort low; Reasoning level: Low44.9%not datedcursor.com
40claude-opus-5_highAnthropicclosedeffort high; Reasoning level: High44.7%not datedcursor.com
41claude-haiku-5-5_xhighAnthropicclosedeffort xhigh; Reasoning level: Extra High44.3%not datedcursor.com
42grok-4.7_highxAIclosedeffort high; Reasoning level: High43.9%not datedcursor.com
43claude-opus-5-5_lowAnthropicclosedeffort low; Reasoning level: Low43.7%not datedcursor.com
44claude-opus-5_mediumAnthropicclosedeffort medium; Reasoning level: Medium43.3%not datedcursor.com
45zai-org/GLM-5.3Epoch: glm-5.3_max, Z.ai (Zhipu AI)openeffort max; Reasoning level: Max42.6%not datedcursor.com
46claude-haiku-5-5_highAnthropicclosedeffort high; Reasoning level: High42.3%not datedcursor.com
47gpt-5.6-sol_maxOpenAIclosedeffort max; Reasoning level: Max41.7%not datedcursor.com
48muse-wonder-1.0_lowMeta AIclosedeffort low; Reasoning level: Low41.7%not datedcursor.com
49grok-4.7_mediumxAIclosedeffort medium; Reasoning level: Medium41.6%not datedcursor.com
50muse-spark-1.3_maxMeta AIclosedeffort max; Reasoning level: Max41.6%not datedcursor.com
51grok-4.6_xhighxAIclosedeffort xhigh; Reasoning level: Extra High41.4%not datedcursor.com
52gpt-5.6-terra_maxOpenAIclosedeffort max; Reasoning level: Max41.3%not datedcursor.com
53claude-opus-5_lowAnthropicclosedeffort low; Reasoning level: Low40.7%not datedcursor.com
54grok-4.6_highxAIclosedeffort high; Reasoning level: High40.4%not datedcursor.com
55gemini-3.8-flash_highGoogle DeepMindclosedeffort high; Reasoning level: High39.6%not datedcursor.com
56claude-sonnet-5-5_mediumAnthropicclosedeffort medium; Reasoning level: Medium39.2%not datedcursor.com
57zai-org/GLM-5.3Epoch: glm-5.3_high, Z.ai (Zhipu AI)openeffort high; Reasoning level: High38.0%not datedcursor.com
58gpt-5.6-sol_xhighOpenAIclosedeffort xhigh; Reasoning level: Extra High37.7%not datedcursor.com
59muse-spark-1.3_xhighMeta AIclosedeffort xhigh; Reasoning level: Extra High37.5%not datedcursor.com
60gemini-3.8-flash_mediumGoogle DeepMindclosedeffort medium; Reasoning level: Medium37.3%not datedcursor.com
61claude-haiku-5-5_mediumAnthropicclosedeffort medium; Reasoning level: Medium36.9%not datedcursor.com
62zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_max, Z.ai (Zhipu AI)openeffort max; Reasoning level: Max36.8%not datedcursor.com
63grok-4.6_mediumxAIclosedeffort medium; Reasoning level: Medium36.1%not datedcursor.com
64gpt-5.6-luna_maxOpenAIclosedeffort max; Reasoning level: Max35.9%not datedcursor.com
65claude-sonnet-5-5_lowAnthropicclosedeffort low; Reasoning level: Low35.8%not datedcursor.com
66gpt-5.6-sol_highOpenAIclosedeffort high; Reasoning level: High35.7%not datedcursor.com
67muse-wonder-1.0_minimalMeta AIclosedeffort minimal34.2%not datedcursor.com
68claude-sonnet-5_maxAnthropicclosedeffort max; Reasoning level: Max34.1%not datedcursor.com
69gpt-5.6-terra_xhighOpenAIclosedeffort xhigh; Reasoning level: Extra High33.6%not datedcursor.com
70grok-4.6_lowxAIclosedeffort low; Reasoning level: Low33.4%not datedcursor.com
71muse-spark-1.3_highMeta AIclosedeffort high; Reasoning level: High33.4%not datedcursor.com
72zai-org/GLM-5.3Epoch: glm-5.3_low, Z.ai (Zhipu AI)openeffort low; Reasoning level: Low33.3%not datedcursor.com
73grok-4.7_lowxAIclosedeffort low; Reasoning level: Low33.1%not datedcursor.com
74gpt-5.6-luna_xhighOpenAIclosedeffort xhigh; Reasoning level: Extra High33.0%not datedcursor.com
75muse-spark-1.3_mediumMeta AIclosedeffort medium; Reasoning level: Medium32.6%not datedcursor.com
76claude-sonnet-5_xhighAnthropicclosedeffort xhigh; Reasoning level: Extra High32.0%not datedcursor.com
77zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_high, Z.ai (Zhipu AI)openeffort high; Reasoning level: High31.1%not datedcursor.com
78gpt-5.6-sol_mediumOpenAIclosedeffort medium; Reasoning level: Medium31.1%not datedcursor.com
79claude-haiku-5-5_lowAnthropicclosedeffort low; Reasoning level: Low30.9%not datedcursor.com
80claude-sonnet-5_highAnthropicclosedeffort high; Reasoning level: High30.8%not datedcursor.com
81gpt-5.6-terra_highOpenAIclosedeffort high; Reasoning level: High30.7%not datedcursor.com
82gpt-5.6-luna_highOpenAIclosedeffort high; Reasoning level: High29.4%not datedcursor.com
83muse-spark-1.3_lowMeta AIclosedeffort low; Reasoning level: Low29.3%not datedcursor.com
84claude-sonnet-5_mediumAnthropicclosedeffort medium; Reasoning level: Medium28.0%not datedcursor.com
85gemini-4-hydrogen_lowGoogle DeepMindclosedeffort low; Reasoning level: Low27.9%not datedcursor.com
86Composer 2.5Cursorcloseddefault27.7%not datedcursor.com
87gpt-5.6-terra_mediumOpenAIclosedeffort medium; Reasoning level: Medium27.6%not datedcursor.com
88zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_low, Z.ai (Zhipu AI)openeffort low; Reasoning level: Low26.9%not datedcursor.com
89gpt-5.6-terra_lowOpenAIclosedeffort low; Reasoning level: Low25.2%not datedcursor.com
90gpt-5.6-sol_lowOpenAIclosedeffort low; Reasoning level: Low24.6%not datedcursor.com
91muse-spark-1.3_minimalMeta AIclosedeffort minimal24.3%not datedcursor.com
92claude-sonnet-5_lowAnthropicclosedeffort low; Reasoning level: Low24.1%not datedcursor.com
93gpt-5.6-luna_mediumOpenAIclosedeffort medium; Reasoning level: Medium22.2%not datedcursor.com
94gpt-5.6-luna_lowOpenAIclosedeffort low; Reasoning level: Low16.0%not datedcursor.com

Open-weight rows are shaded. Dates are when the result was run or evaluated, where the data says. Data from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Epoch AI. Method