Benchmark Almanac

CursorBench 4.1

Gemini 4 Neon 65.4% leads since November 2026
In use
Version 4.1

Ambiguous multi-file coding tasks from real Cursor sessions in Cursor's production harness. Agentic graders score correctness, with cost reported.

Google first reports version 4.1 in its Gemini 4 Neon launch post. Cursor has not said what changed from 4.0, so scores may not compare.

Made by
Cursor
Released
November 2026
Kind
Fixing real repositories
Level
Expert
Scored
Graded by a model
Answers
Code
Tasks
Kept private
Measures
Capability
Language
English
In and out
Text, Code → Code, Actions
Interaction
Agentic
Environment
Terminal
Items from
Real work and use
Run by
A host
Based on
CursorBench 4.0
In views
Agents
Strongest model tested
Claude Fable 5.2 #2 on the capability index
Last in a launch post
November 2026
Last model added
November 2026
Scores refreshed
November 20th, 2026
12
Models scored on its own leaderboard
1
Launch post lists it, from 1 lab
1
System card reports it, from 1 lab

Gemini 4 Neon leads with 65.4%

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.

  1. 1 Gemini 4 Neon xhigh 65.4%
  2. 2 Gemini 4 Carbon high 61.5%
  3. 3 Gemini 4 Hydrogen xhigh 61.4%
  4. 4 Claude Fable 5.2 max 61.3%
  5. 5 Claude Opus 5.5 max 58.1%
  6. 6 Gemini 4 Argon high 56.5%
  7. 7 Claude Sonnet 5.5 max 54.4%
  8. 8 Grok 4.7 xhigh 48.6%
  9. 9 Claude Haiku 5.5 max 48.4%
  10. 10 Grok 4.6 xhigh 44.3%
  11. 11 Muse Spark 1.3 max 41.9%
  12. 12 GPT-5.6 Sol max 40.9%

Source: Official CursorBench 4.1 leaderboard

#ModelScoreSettingDateSource
1 Gemini 4 Neon 65.4% xhigh, Cursor Nov 2026 Official leaderboard
2 Gemini 4 Carbon 61.5% high, Cursor Oct 2026 Official leaderboard
3 Gemini 4 Hydrogen 61.4% xhigh, Cursor Nov 2026 Official leaderboard
4 Claude Fable 5.2 61.3% max, Cursor Nov 2026 Official leaderboard
5 Claude Opus 5.5 58.1% max, Cursor Sep 2026 Official leaderboard
6 Gemini 4 Argon 56.5% high, Cursor Sep 2026 Official leaderboard
7 Claude Sonnet 5.5 54.4% max, Cursor Sep 2026 Official leaderboard
8 Grok 4.7 48.6% xhigh, Cursor Sep 2026 Official leaderboard
9 Claude Haiku 5.5 48.4% max, Cursor Oct 2026 Official leaderboard
10 Grok 4.6 44.3% xhigh, Cursor Aug 2026 Official leaderboard
11 Muse Spark 1.3 41.9% max, Cursor Sep 2026 Official leaderboard
12 GPT-5.6 Sol 40.9% max, Cursor Jul 2026 Official leaderboard

Labs report scores for 1 model in their cards. We plot every source in the charts below.

CursorBench score by average cost per task

The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

0% 20% 40% 60% 80% $0$10$20$30$40$50 Average cost per task ($), lower is better CursorBench score Gemini 4 Neon (xhigh)Gemini 4 Neon (high)Gemini 4 Neon (medium)Claude Opus 5.5 (medium)Claude Haiku 5.5 (max)Claude Haiku 5.5 (xhigh)Claude Haiku 5.5 (high)Claude Haiku 5.5 (medium)Claude Haiku 5.5 (low) 0% 20% 40% 60% 80% $0$10$20$30$40$50 Average cost per task ($), lower is better CursorBench score Gemini 4 Neon (xhigh)Gemini 4 Neon (high)Gemini 4 Neon (medium)Claude Opus 5.5 (medium)Claude Haiku 5.5 (max)Claude Haiku 5.5 (xhigh)Claude Haiku 5.5 (high)Claude Haiku 5.5 (medium)Claude Haiku 5.5 (low)
ModelSettingCursorBench scoreAverage cost per task
Gemini 4 Neon xhigh, Cursor 65.4% $10.67
Gemini 4 Neon high, Cursor 63.7% $6.54
Gemini 4 Carbon high, Cursor 61.5% $8.52
Gemini 4 Hydrogen xhigh, Cursor 61.4% $8.09
Claude Fable 5.2 max, Cursor 61.3% $48.11
Gemini 4 Carbon medium, Cursor 60.4% $7.22
Claude Fable 5.2 xhigh, Cursor 58.8% $17.84
Claude Opus 5.5 max, Cursor 58.1% $11.88
Gemini 4 Neon medium, Cursor 57.7% $3.07
Gemini 4 Hydrogen high, Cursor 57.3% $4.57
Claude Opus 5.5 xhigh, Cursor 56.7% $6.54
Gemini 4 Argon high, Cursor 56.5% $9.14
Claude Fable 5.2 high, Cursor 56.5% $10.51
Gemini 4 Argon medium, Cursor 55.8% $7.64
Gemini 4 Carbon low, Cursor 55% $8.64
Claude Opus 5.5 high, Cursor 54.5% $3.87
Claude Sonnet 5.5 max, Cursor 54.4% $8.63
Claude Fable 5.2 medium, Cursor 54.2% $5.01
Claude Fable 5.2 low, Cursor 53.4% $4.26
Claude Opus 5.5 medium, Cursor 50% $2.85
Claude Sonnet 5.5 xhigh, Cursor 49.9% $3.13
Grok 4.7 xhigh, Cursor 48.6% $10.19
Claude Haiku 5.5 max, Cursor 48.4% $1.14
Gemini 4 Argon low, Cursor 47.9% $7.96
Gemini 4 Hydrogen medium, Cursor 46.4% $1.28
Grok 4.7 high, Cursor 46% $6.88
Gemini 4 Neon low, Cursor 45.5% $1.26
Claude Haiku 5.5 xhigh, Cursor 45.2% $0.56
Grok 4.6 xhigh, Cursor 44.3% $5.26
Claude Sonnet 5.5 high, Cursor 43.8% $1.41
Claude Opus 5.5 low, Cursor 42.7% $1.43
Grok 4.7 medium, Cursor 42.5% $5.52
Muse Spark 1.3 max, Cursor 41.9% $2.89
Grok 4.6 high, Cursor 41.3% $4.28
GPT-5.6 Sol max, Cursor 40.9% $7.6
Grok 4.6 medium, Cursor 38.9% $2.59
Claude Haiku 5.5 high, Cursor 37.5% $0.31
GPT-5.6 Sol xhigh, Cursor 37.4% $4.32
Muse Spark 1.3 xhigh, Cursor 34.6% $2.33
Claude Sonnet 5.5 medium, Cursor 34.3% $0.63
Claude Haiku 5.5 medium, Cursor 34.2% $0.17
GPT-5.6 Sol high, Cursor 33.2% $3
Grok 4.6 low, Cursor 33.1% $1.76
Muse Spark 1.3 high, Cursor 32.1% $1.97
Claude Sonnet 5.5 low, Cursor 31.8% $0.44
Grok 4.7 low, Cursor 31.5% $2.73
Muse Spark 1.3 medium, Cursor 31.3% $1.75
GPT-5.6 Sol medium, Cursor 28.7% $1.97
Claude Haiku 5.5 low, Cursor 27.8% $0.08
Muse Spark 1.3 low, Cursor 27.5% $1.13
Gemini 4 Hydrogen low, Cursor 23.9% $0.28
GPT-5.6 Sol low, Cursor 23% $0.98
Muse Spark 1.3 minimal, Cursor 21.3% $0.69

The best score, 65.4%, costs $11 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Anthropic Google DeepMind Cost frontier One model at each effort level
0% 20% 40% 60% 80% $0.01 $0.10 $1.00 $10 $100 Cost per task, log scale Gemini 4 Neon, xhigh Gemini 4 Neon, medium Claude Opus 5.5, medium Claude Haiku 5.5, xhigh Claude Haiku 5.5, high Claude Haiku 5.5, medium Claude Haiku 5.5, low 0% 20% 40% 60% 80% $0.01 $0.10 $1.00 $10 $100 Cost per task, log scale Gemini 4 Neon, xhigh Gemini 4 Neon, medium

Source: Official CursorBench 4.1 leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 4 Neon xhigh Cursor 65.4% $11 136,273
Gemini 4 Neon high Cursor 63.7% $6.54 88,596
Gemini 4 Carbon high Cursor 61.5% $8.52 114,518
Gemini 4 Hydrogen xhigh Cursor 61.4% $8.09 205,981
Claude Fable 5.2 max Cursor 61.3% $48 541,003
Gemini 4 Carbon medium Cursor 60.4% $7.22 103,189
Claude Fable 5.2 xhigh Cursor 58.8% $18 194,918
Claude Opus 5.5 max Cursor 58.1% $12 213,070
Gemini 4 Neon medium Cursor 57.7% $3.07 47,059
Gemini 4 Hydrogen high Cursor 57.3% $4.57 113,807
Claude Opus 5.5 xhigh Cursor 56.7% $6.54 104,901
Gemini 4 Argon high Cursor 56.5% $9.14 135,038
Claude Fable 5.2 high Cursor 56.5% $11 108,213
Gemini 4 Argon medium Cursor 55.8% $7.64 128,231
Gemini 4 Carbon low Cursor 55% $8.64 48,526
Claude Opus 5.5 high Cursor 54.5% $3.87 56,162
Claude Sonnet 5.5 max Cursor 54.4% $8.63 342,273
Claude Fable 5.2 medium Cursor 54.2% $5.01 47,678
Claude Fable 5.2 low Cursor 53.4% $4.26 40,384
Claude Opus 5.5 medium Cursor 50% $2.85 39,736
Claude Sonnet 5.5 xhigh Cursor 49.9% $3.13 112,299
Grok 4.7 xhigh Cursor 48.6% $10 90,268
Claude Haiku 5.5 max Cursor 48.4% $1.14 344,890
Gemini 4 Argon low Cursor 47.9% $7.96 40,541
Gemini 4 Hydrogen medium Cursor 46.4% $1.28 43,606
Grok 4.7 high Cursor 46% $6.88 65,850
Gemini 4 Neon low Cursor 45.5% $1.26 19,580
Claude Haiku 5.5 xhigh Cursor 45.2% $0.56 149,314
Grok 4.6 xhigh Cursor 44.3% $5.26 58,554
Claude Sonnet 5.5 high Cursor 43.8% $1.41 43,088
Claude Opus 5.5 low Cursor 42.7% $1.43 19,497
Grok 4.7 medium Cursor 42.5% $5.52 41,173
Muse Spark 1.3 max Cursor 41.9% $2.89 56,572
Grok 4.6 high Cursor 41.3% $4.28 46,330
GPT-5.6 Sol max Cursor 40.9% $7.60 40,466
Grok 4.6 medium Cursor 38.9% $2.59 26,374
Claude Haiku 5.5 high Cursor 37.5% $0.31 77,196
GPT-5.6 Sol xhigh Cursor 37.4% $4.32 23,921
Muse Spark 1.3 xhigh Cursor 34.6% $2.33 47,142
Claude Sonnet 5.5 medium Cursor 34.3% $0.63 19,144
Claude Haiku 5.5 medium Cursor 34.2% $0.17 43,441
GPT-5.6 Sol high Cursor 33.2% $3.00 16,319
Grok 4.6 low Cursor 33.1% $1.76 16,863
Muse Spark 1.3 high Cursor 32.1% $1.97 36,562
Claude Sonnet 5.5 low Cursor 31.8% $0.44 13,798
Grok 4.7 low Cursor 31.5% $2.73 14,493
Muse Spark 1.3 medium Cursor 31.3% $1.75 32,034
GPT-5.6 Sol medium Cursor 28.7% $1.97 10,554
Claude Haiku 5.5 low Cursor 27.8% $0.08 23,673
Muse Spark 1.3 low Cursor 27.5% $1.13 20,673
Gemini 4 Hydrogen low Cursor 23.9% $0.28 10,497
GPT-5.6 Sol low Cursor 23% $0.98 5,206
Muse Spark 1.3 minimal Cursor 21.3% $0.69 12,789

2 models have held number one on the official leaderboard since its release in November 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

Google DeepMind
Own leaderboard 2 leaders

Source: Official CursorBench 4.1 leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2027 Near the ceiling (85%) Gemini 4 Neon Claude Opus 5.5 Grok 4.7 Grok 4.6 0% 20% 40% 60% 80% 100% 2027 Near the ceiling (85%) Gemini 4 Neon Claude Opus 5.5

Sources: Official CursorBench 4.1 leaderboard and labs’ model cards

One lab put it in a launch post in the last six months

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out. Anthropic accounts for 7 of the 14 releases that named it since 2024.

Google DeepMind
DateLaunch postScoreAs the post writes it
Nov 18, 2026 Gemini 4 Neon 65.4 CursorBench 4.1 (Score)

One system card reports it

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
Nov 18, 2026 Gemini 4 Neon model card 65.4 CursorBench 4.1 (Score) (results table, p.4) Capabilities

One lab claim next to the leaderboard

Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. 1 of them match, which usually means one side copied the other. Every benchmark.

Model Lab says Board says Gap
Gemini 4 Neon 65.4% CursorBench 4.1 (Score) 65.4% Leaderboard, xhigh, Cursor Same

Versions

We switch the top row to each new version when it comes out. Open it for one row per version. An old version stays in play while people still test it, so whoever first takes it past 85% gets credit for closing it out.

Anthropic xAI Google DeepMind
CursorBench 4.1 11 leaders
CursorBench 3.1 Replaced by 3.2, Jul 2026
CursorBench 3.2 Replaced by 4.0, Sep 2026
CursorBench 4.0 5 leaders
CursorBench 4.1 2 leaders

Sources: Official leaderboards and labs’ model cards

More in Coding

All 523 in Coding
  • SWE-bench Verified In use 500 human-validated GitHub issues from 12 Python repos. Agent writes a patch that must pass hidden tests. Claude Opus 4.7 83.5%
  • LiveCodeBench v6 In use Competitive programming problems collected after model cutoffs from LeetCode, AtCoder, Codeforces. Scored pass@1 on hidden tests. DeepSeek-V4-Pro 93.5%
  • SWE-bench Multilingual In use 300 GitHub-issue tasks across nine programming languages in SWE-bench format. Patch graded by repository tests. Claude Fable 5.2 96.6%
  • DeepSWE In use 113 original long-horizon engineering tasks across 91 repos and five languages, written from scratch. Hand-written verifiers accept alternative solutions. GPT-6 Astra 74.1%
  • SciCode In use Scientist-curated research coding problems split into subproblems across natural sciences. Code checked against test cases. Seed2.1 Pro 59.8%
  • Codeforces rating In use Model solves recent Codeforces contest problems. An Elo-style rating is computed from simulated contest performance. DeepSeek V4.1 Flash 3,471

Data audit

11 of 12 checks pass, 1 warning. Look first at the Confidence check. Likely. Some facts rest on secondary sources. Every benchmark

Catalog record 1 warning 1 pass
  • Pass Task access The tasks are private.
  • Warning Confidence Likely. Some facts rest on secondary sources.
Scores 5 pass
  • Pass Independent runs Scores from the official leaderboard (12 entries), launch posts (1 score) and system cards (1 score).
  • Pass Models scored 12 models we track.
  • Pass Newest score The newest is from November 2026, for Gemini 4 Neon.
  • Pass Frontier models 5 of the top 10 have a score, led by Claude Fable 5.2 (#2).
  • Pass Headroom The best is 65.4%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on cursor.com: 53 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 2 pass
  • Pass Lab claims The one pair agrees within 10 points.
  • Pass Known issues No published audit found problems.

How we check