CursorBench 3.2, Grok 4.6 High (xAI first-party evals table)
Observed Aug 12, 2026 · 1 configuration.
#Published configuration and provenanceRelative scoreScore
activeCodingAgents
CursorBench
Metric: Score (higher is better)Introduced: 2026
This page is LLMReference's October 5, 2026 historical snapshot of CursorBench 4.0, not the current CursorBench release. Use it to compare Cursor IDE-native coding-agent configurations for multi-file work when Cursor is the target Coding or Agents workflow; it does not establish a neutral base-model winner. Read each score with that configuration's cost, token, step, and qualification context.
Configurations
46
tracked in CursorBench 4.0
Score band
57.8 – 16.0
best → lowest tracked
Snapshot trend
+17.50
CursorBench 4.0 · Sep 10 → Oct 5 · 3 models
Published snapshot
CursorBench 4.0 · observed Oct 5, 2026 · 46 configurations.
CursorBench 3.2 configurations are published exactly as Cursor reports them, including cost, token, and step averages. No per-model maximum is selected. Grok 4.5 rows carry Cursor's training-data-advantage disclosure and are not neutral winner evidence.
#Published configuration and provenanceRelative scoreScore
1
Claude Opus 5.5
57.8Configuration: claude-opus-5-5 (Max)
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Oct 5, 2026Confidence: confirmedCost/task: $13.43Tokens/task: 218,363Steps/task: 185Source
Notes: Cursor vendor-reported CursorBench 4.0 configuration result; verified on https://cursor.com/cursorbench chart label 'Opus 5.5 Max' (retrieved 2026-10-05). Not independently reproducible. Highest-scoring published configuration selected; timestamp is retrieval date (page carries no per-row date). Cost/tokens/steps from cursor.com/cursorbench table row (same retrieval).
2
Claude Sonnet 5.5
55.5Configuration: claude-sonnet-5-5 (Max)
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Oct 5, 2026Confidence: confirmedCost/task: $9.67Tokens/task: 271,920Steps/task: 170Source
Notes: Cursor vendor-reported CursorBench 4.0 configuration result; verified on https://cursor.com/cursorbench chart label 'Sonnet 5.5 Max' (retrieved 2026-10-05). Not independently reproducible. Highest-scoring published configuration selected; timestamp is retrieval date (page carries no per-row date). Cost/tokens/steps from cursor.com/cursorbench table row (same retrieval).
3
Claude Fable 5.1
51.8Configuration: Fable 5.1 Max
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $17.28Tokens/task: 117,236Steps/task: 128Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
4
Claude Fable 5.1
51.6Configuration: Fable 5.1 Extra High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $13.01Tokens/task: 87,294Steps/task: 101Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
5
Claude Fable 5.1
49.2Configuration: Fable 5.1 High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $9.08Tokens/task: 58,438Steps/task: 77Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
6
Claude Fable 5.1
46.8Configuration: Fable 5.1 Medium
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $7.05Tokens/task: 45,411Steps/task: 63Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
7
Claude Opus 5
46.6Configuration: Opus 5 Max
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $11.95Tokens/task: 85,384Steps/task: 106Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
8
Grok 4.7
46.3Configuration: grok-4.7 (Extra High)
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Oct 5, 2026Confidence: confirmedCost/task: $6.01Tokens/task: 70,141Steps/task: 88Source
Notes: Cursor vendor-reported CursorBench 4.0 configuration result; verified on https://cursor.com/cursorbench chart label 'Grok 4.7 Extra High' (retrieved 2026-10-05). Not independently reproducible. Highest-scoring published configuration selected; timestamp is retrieval date (page carries no per-row date). Cost/tokens/steps from cursor.com/cursorbench table row (same retrieval).
9
Claude Opus 5
46.1Configuration: Opus 5 Extra High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $11.43Tokens/task: 80,094Steps/task: 103Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
10
Claude Fable 5.1
45.1Configuration: Fable 5.1 Low
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $5.44Tokens/task: 34,795Steps/task: 51Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
11
Claude Opus 5
44.7Configuration: Opus 5 High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $9.00Tokens/task: 61,405Steps/task: 86Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
12
Claude Opus 5
43.3Configuration: Opus 5 Medium
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $6.94Tokens/task: 45,272Steps/task: 72Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
13
GPT-5.6 Sol
41.7Configuration: GPT-5.6 Sol Max
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $8.23Tokens/task: 42,944Steps/task: 99Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
14
Muse Spark 1.3
41.6Configuration: Muse Spark 1.3 Max
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $2.64Tokens/task: 52,005Steps/task: 98Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
15
Grok 4.6
41.4Configuration: Grok 4.6 Extra High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $6.10Tokens/task: 49,814Steps/task: 56Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
16
GPT-5.6 Terra
41.3Configuration: GPT-5.6 Terra Max
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $5.14Tokens/task: 60,814Steps/task: 107Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
17
Claude Opus 5
40.7Configuration: Opus 5 Low
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $4.87Tokens/task: 31,995Steps/task: 57Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
18
Grok 4.6
40.4Configuration: Grok 4.6 High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $5.20Tokens/task: 41,387Steps/task: 48Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
19
Gemini 3.8 Flash
39.6Configuration: Gemini 3.8 Flash High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $4.70Tokens/task: 162,565Steps/task: 324Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
20
GPT-5.6 Sol
37.7Configuration: GPT-5.6 Sol Extra High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $4.40Tokens/task: 24,729Steps/task: 55Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
21
Muse Spark 1.3
37.5Configuration: Muse Spark 1.3 Extra High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $2.10Tokens/task: 40,891Steps/task: 83Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
22
Gemini 3.8 Flash
37.3Configuration: Gemini 3.8 Flash Medium
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $4.06Tokens/task: 128,364Steps/task: 290Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
23
Grok 4.6
36.1Configuration: Grok 4.6 Medium
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $3.48Tokens/task: 24,893Steps/task: 40Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
24
GPT-5.6 Luna
35.9Configuration: GPT-5.6 Luna Max
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $1.03Tokens/task: 87,284Steps/task: 208Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
25
GPT-5.6 Sol
35.7Configuration: GPT-5.6 Sol High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $2.85Tokens/task: 16,174Steps/task: 41Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
26
Claude Sonnet 5
34.1Configuration: Sonnet 5 Max
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $7.17Tokens/task: 149,257Steps/task: 140Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
27
GPT-5.6 Terra
33.6Configuration: GPT-5.6 Terra Extra High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $1.81Tokens/task: 23,436Steps/task: 43Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
28
Grok 4.6
33.4Configuration: Grok 4.6 Low
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $2.25Tokens/task: 16,307Steps/task: 32Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
29
Muse Spark 1.3
33.4Configuration: Muse Spark 1.3 High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $1.66Tokens/task: 30,654Steps/task: 69Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
30
GPT-5.6 Luna
33.0Configuration: GPT-5.6 Luna Extra High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $0.44Tokens/task: 40,598Steps/task: 98Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
31
Muse Spark 1.3
32.6Configuration: Muse Spark 1.3 Medium
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $1.49Tokens/task: 27,255Steps/task: 64Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
32
Claude Sonnet 5
32.0Configuration: Sonnet 5 Extra High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $4.55Tokens/task: 83,373Steps/task: 102Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
33
GPT-5.6 Sol
31.1Configuration: GPT-5.6 Sol Medium
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $1.77Tokens/task: 10,111Steps/task: 32Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
34
Claude Sonnet 5
30.8Configuration: Sonnet 5 High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $3.48Tokens/task: 61,146Steps/task: 85Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
35
GPT-5.6 Terra
30.7Configuration: GPT-5.6 Terra High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $1.11Tokens/task: 13,162Steps/task: 33Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
36
GPT-5.6 Luna
29.4Configuration: GPT-5.6 Luna High
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $0.25Tokens/task: 23,368Steps/task: 64Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
37
Muse Spark 1.3
29.3Configuration: Muse Spark 1.3 Low
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $0.93Tokens/task: 17,483Steps/task: 47Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
38
Claude Sonnet 5
28.0Configuration: Sonnet 5 Medium
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $2.31Tokens/task: 39,114Steps/task: 65Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
39
Composer 2.5
27.7Configuration: Composer 2.5
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $0.68Tokens/task: 17,347Steps/task: 41Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
40
GPT-5.6 Terra
27.6Configuration: GPT-5.6 Terra Medium
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $0.64Tokens/task: 7,307Steps/task: 25Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
41
GPT-5.6 Terra
25.2Configuration: GPT-5.6 Terra Low
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $0.52Tokens/task: 5,914Steps/task: 23Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
42
GPT-5.6 Sol
24.6Configuration: GPT-5.6 Sol Low
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $0.87Tokens/task: 4,885Steps/task: 21Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
43
Muse Spark 1.3
24.3Configuration: Muse Spark 1.3 Minimal
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $0.56Tokens/task: 10,620Steps/task: 34Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
44
Claude Sonnet 5
24.1Configuration: Sonnet 5 Low
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $1.39Tokens/task: 23,772Steps/task: 46Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
45
GPT-5.6 Luna
22.2Configuration: GPT-5.6 Luna Medium
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $0.08Tokens/task: 7,642Steps/task: 32Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
46
GPT-5.6 Luna
16.0Configuration: GPT-5.6 Luna Low
Version: CursorBench 4.0Harness: CursorBench 4.0 productized Cursor-agent workflowEvaluator: CursorObserved: Sep 10, 2026Confidence: confirmedCost/task: $0.03Tokens/task: 3,288Steps/task: 18Source
Notes: Cursor vendor-reported configuration result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
Other version snapshots
Every benchmark version stays in a separate dated table. Scores, ranks, and changes are not compared across versions.
CursorBench 3.2
Observed Jul 18, 2026 · 42 configurations.
#Published configuration and provenanceRelative scoreScore
1
Claude Fable 5
70.5Configuration: Fable 5 Max
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $17.32Tokens/task: 103,525Steps/task: 72Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
2
Claude Fable 5
68.4Configuration: Fable 5 Extra High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $11.73Tokens/task: 64,971Steps/task: 56Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
3
GPT-5.6 Sol
67.2Configuration: GPT-5.6 Sol Max
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $5.69Tokens/task: 28,320Steps/task: 48Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
4
Grok 4.5
66.7Configuration: Grok 4.5 High*
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.51Tokens/task: 19,521Steps/task: 33Source
Qualification: Cursor discloses a training-data advantage; do not use this result for a neutral ranking claim.
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
5
Claude Fable 5
66.5Configuration: Fable 5 High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $8.77Tokens/task: 43,747Steps/task: 48Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
6
Grok 4.5
65.4Configuration: Grok 4.5 Medium*
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.54Tokens/task: 18,914Steps/task: 34Source
Qualification: Cursor discloses a training-data advantage; do not use this result for a neutral ranking claim.
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
7
Claude Fable 5
65.2Configuration: Fable 5 Medium
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $6.80Tokens/task: 30,366Steps/task: 41Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
8
GPT-5.6 Terra
64.9Configuration: GPT-5.6 Terra Max
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $2.89Tokens/task: 32,969Steps/task: 47Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
9
GPT-5.6 Sol
64.5Configuration: GPT-5.6 Sol Extra High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $3.88Tokens/task: 19,699Steps/task: 38Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
10
Grok 4.5
63.5Configuration: Grok 4.5 Low*
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.22Tokens/task: 15,841Steps/task: 31Source
Qualification: Cursor discloses a training-data advantage; do not use this result for a neutral ranking claim.
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
11
GPT-5.6 Sol
63.5Configuration: GPT-5.6 Sol High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $2.79Tokens/task: 13,867Steps/task: 32Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
12
Claude Opus 4.8
62.3Configuration: Opus 4.8 Max
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $5.77Tokens/task: 71,411Steps/task: 44Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
13
Claude Fable 5
62.1Configuration: Fable 5 Low
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $4.46Tokens/task: 18,182Steps/task: 31Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
14
Claude Sonnet 5
61.5Configuration: Sonnet 5 Max
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $6.45Tokens/task: 92,882Steps/task: 86Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
15
GPT-5.6 Luna
61.1Configuration: GPT-5.6 Luna Max
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.97Tokens/task: 87,973Steps/task: 61Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
16
GPT-5.6 Sol
60.0Configuration: GPT-5.6 Sol Medium
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.95Tokens/task: 9,747Steps/task: 27Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
17
Claude Opus 4.8
59.4Configuration: Opus 4.8 Extra High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $4.50Tokens/task: 51,121Steps/task: 40Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
18
GPT-5.6 Terra
59.2Configuration: GPT-5.6 Terra Extra High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.44Tokens/task: 16,089Steps/task: 29Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
19
Claude Sonnet 5
58.7Configuration: Sonnet 5 Extra High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $4.16Tokens/task: 52,871Steps/task: 67Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
20
GPT-5.5
58.4Configuration: GPT-5.5 High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $2.05Tokens/task: 12,183Steps/task: 28Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
21
GPT-5.5
58.4Configuration: GPT-5.5 Extra High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $2.85Tokens/task: 17,534Steps/task: 32Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
22
Claude Opus 4.8
58.0Configuration: Opus 4.8 High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $3.15Tokens/task: 33,548Steps/task: 33Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
23
GPT-5.6 Luna
57.7Configuration: GPT-5.6 Luna Extra High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.14Tokens/task: 22,480Steps/task: 48Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
24
Claude Sonnet 5
56.9Configuration: Sonnet 5 High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $3.19Tokens/task: 39,483Steps/task: 57Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
25
GPT-5.6 Luna
56.8Configuration: GPT-5.6 Luna High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $0.82Tokens/task: 15,141Steps/task: 40Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
26
Claude Opus 4.8
56.1Configuration: Opus 4.8 Medium
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $2.81Tokens/task: 28,384Steps/task: 32Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
27
Composer 2.5
56.1Configuration: Composer 2.5
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $0.44Tokens/task: 14,286Steps/task: 33Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
28
GLM-5.2
55.0Configuration: GLM 5.2 Max
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.76Tokens/task: 35,946Steps/task: 58Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
29
GPT-5.6 Terra
54.2Configuration: GPT-5.6 Terra High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $0.89Tokens/task: 9,468Steps/task: 23Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
30
GPT-5.5
53.8Configuration: GPT-5.5 Medium
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.51Tokens/task: 8,522Steps/task: 25Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
31
Claude Opus 4.8
53.1Configuration: Opus 4.8 Low
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $2.02Tokens/task: 19,624Steps/task: 27Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
32
GPT-5.6 Sol
52.6Configuration: GPT-5.6 Sol Low
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.01Tokens/task: 5,104Steps/task: 19Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
33
Claude Sonnet 5
52.4Configuration: Sonnet 5 Medium
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $2.16Tokens/task: 26,200Steps/task: 46Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
34
GLM-5.2
51.5Configuration: GLM 5.2 High
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.19Tokens/task: 21,829Steps/task: 49Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
35
GPT-5.6 Terra
50.3Configuration: GPT-5.6 Terra Medium
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $0.61Tokens/task: 6,222Steps/task: 20Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
36
Kimi K2.7-Code
49.7Configuration: Kimi K2.7 Code
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.43Tokens/task: 31,247Steps/task: 58Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
37
Gemini 3.5 Flash
48.8Configuration: Gemini 3.5 Flash
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $2.20Tokens/task: 46,702Steps/task: 77Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
38
GPT-5.6 Luna
47.7Configuration: GPT-5.6 Luna Medium
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $0.39Tokens/task: 7,095Steps/task: 28Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
39
Claude Sonnet 5
47.7Configuration: Sonnet 5 Low
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $1.30Tokens/task: 16,269Steps/task: 33Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
40
GPT-5.6 Terra
46.9Configuration: GPT-5.6 Terra Low
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $0.53Tokens/task: 5,312Steps/task: 19Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
41
GPT-5.5
46.6Configuration: GPT-5.5 Low
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $0.98Tokens/task: 5,168Steps/task: 20Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
42
GPT-5.6 Luna
37.6Configuration: GPT-5.6 Luna Low
Version: CursorBench 3.2Harness: CursorBench 3.2 productized Cursor-agent workflowEvaluator: CursorObserved: Jul 18, 2026Confidence: confirmedCost/task: $0.16Tokens/task: 3,209Steps/task: 17Source
Notes: Cursor vendor-reported result; not independently reproducible. Results are subject to variance, and small score differences may not be statistically meaningful.
CursorBench 3.1
Observed Jun 30, 2026 · 12 configurations.
#Published configuration and provenanceRelative scoreScore
1
Claude Fable 5
72.9Configuration: Fable 5 Max
Version: CursorBench 3.1Harness: CursorBench 3.1Evaluator: CursorObserved: Jun 30, 2026Confidence: confirmedSource
Notes: Highest CursorBench 3.1 score across Cursor's published effort configurations for this base model.
2
Claude Opus 4.7
64.8Configuration: Opus 4.7 Max
Version: CursorBench 3.1Harness: CursorBench 3.1Evaluator: CursorObserved: Jun 30, 2026Confidence: confirmedSource
Notes: Highest CursorBench 3.1 score across Cursor's published effort configurations for this base model.
3
GPT-5.5
64.3Configuration: GPT-5.5 Extra High
Version: CursorBench 3.1Harness: CursorBench 3.1Evaluator: CursorObserved: Jun 30, 2026Confidence: confirmedSource
Notes: Highest CursorBench 3.1 score across Cursor's published effort configurations for this base model.
4
Claude Opus 4.8
63.8Configuration: Opus 4.8 Max
Version: CursorBench 3.1Harness: CursorBench 3.1Evaluator: CursorObserved: Jun 30, 2026Confidence: confirmedSource
Notes: Highest CursorBench 3.1 score across Cursor's published effort configurations for this base model.
5
Composer 2.5
63.2Configuration: Composer 2.5 (single reported configuration)
Version: CursorBench 3.1Harness: CursorBench 3.1Evaluator: CursorObserved: Jun 30, 2026Confidence: confirmedSource
Notes: Cursor published one CursorBench 3.1 configuration for this model; no cross-effort selection was needed.
6
Claude Sonnet 5
61.2Configuration: Sonnet 5 Max
Version: CursorBench 3.1Harness: CursorBench 3.1Evaluator: CursorObserved: Jun 30, 2026Confidence: confirmedSource
Notes: Highest CursorBench 3.1 score across Cursor's published effort configurations for this base model.
7
GLM-5.2
54.6Configuration: GLM 5.2 Max
Version: CursorBench 3.1Harness: CursorBench 3.1Evaluator: CursorObserved: Jun 30, 2026Confidence: confirmedSource
Notes: Highest CursorBench 3.1 score across Cursor's published effort configurations for this base model.
8
Composer 2
52.2Configuration: Composer 2 (single reported configuration)
Version: CursorBench 3.1Harness: CursorBench 3.1Evaluator: CursorObserved: Jun 30, 2026Confidence: confirmedSource
Notes: Cursor published one CursorBench 3.1 configuration for this model; no cross-effort selection was needed.
9
Gemini 3.5 Flash
49.8Configuration: Gemini 3.5 Flash (single reported configuration)
Version: CursorBench 3.1Harness: CursorBench 3.1Evaluator: CursorObserved: Jun 30, 2026Confidence: confirmedSource
Notes: Cursor published one CursorBench 3.1 configuration for this model; no cross-effort selection was needed.
10
Claude Sonnet 4.6
49.0Configuration: Sonnet 4.6 Max
Version: CursorBench 3.1Harness: CursorBench 3.1Evaluator: CursorObserved: Jun 30, 2026Confidence: confirmedSource
Notes: Highest CursorBench 3.1 score across Cursor's published effort configurations for this base model.
How to read this benchmark
This benchmark scores models where higher is better. Use scores for directional filtering and shortlisting, not universal quality ranking; then validate pricing, context window, provider availability, and fit for your workload.
Trust this score when
- The row names the exact model configuration, benchmark version, evaluator, harness, and observation date.
- Cost, token, and step averages are evaluated together with the score for the same configuration.
- You validate the configuration in your own Cursor workload before making a model choice.
Be cautious when
- Rows are collapsed into one score per base model or compared across benchmark versions.
- Small score differences are treated as decisive despite Cursor's variance caveat.
- A qualified row, including Grok 4.5's disclosed training-data advantage, is used as neutral winner evidence.
Related benchmarks
Last reviewed: Sep 14, 2026