activeAgentsVision

TuringSpace Pass@1

Metric: Pass@1 (%) (higher is better)Introduced: 2026

TuringSpace evaluates interactive spatial reasoning on 125 programmatic CAPTCHA tasks across five categories. Pass@1 is the mean success rate over test seeds 36, 43 and 50 (375 instances), with a limit of 50 model calls per instance. Pass^3 is a separate metric.

Models ranked

12

tracked on this benchmark

Score band

74.1 – 0.5

best → lowest tracked

Snapshot trend

—

need ≥2 snapshots

Leaderboard

Tracked models ranked by Pass@1 (%) (higher is better).

Compare candidates
#Model variant and provenanceScore
1

Configuration: Pass@1; three seeds 36/43/50; all 125 tasks; mean success rate, not pass^3

Version: TuringSpace-Bench arXiv 2611.25387 v1Harness: Owner public computer-use agents; 125 tasks × test seeds 36,43,50; 1920x1080 screenshots; <=50 model calls per instanceEvaluator: TuringSpace authorsObserved: Nov 20, 2026Confidence: highSource

Notes: Owner-published CSV, pass@1 column only. Human row and models without an unambiguous existing main identity omitted. No model freshness restamp.

74.1
2

Configuration: Pass@1; three seeds 36/43/50; all 125 tasks; mean success rate, not pass^3

Version: TuringSpace-Bench arXiv 2611.25387 v1Harness: Owner public computer-use agents; 125 tasks × test seeds 36,43,50; 1920x1080 screenshots; <=50 model calls per instanceEvaluator: TuringSpace authorsObserved: Nov 20, 2026Confidence: highSource

Notes: Owner-published CSV, pass@1 column only. Human row and models without an unambiguous existing main identity omitted. No model freshness restamp.

65.3
3

Configuration: Pass@1; three seeds 36/43/50; all 125 tasks; mean success rate, not pass^3

Version: TuringSpace-Bench arXiv 2611.25387 v1Harness: Owner public computer-use agents; 125 tasks × test seeds 36,43,50; 1920x1080 screenshots; <=50 model calls per instanceEvaluator: TuringSpace authorsObserved: Nov 20, 2026Confidence: highSource

Notes: Owner-published CSV, pass@1 column only. Human row and models without an unambiguous existing main identity omitted. No model freshness restamp.

57.9
4

Configuration: Pass@1; three seeds 36/43/50; all 125 tasks; mean success rate, not pass^3

Version: TuringSpace-Bench arXiv 2611.25387 v1Harness: Owner public computer-use agents; 125 tasks × test seeds 36,43,50; 1920x1080 screenshots; <=50 model calls per instanceEvaluator: TuringSpace authorsObserved: Nov 20, 2026Confidence: highSource

Notes: Owner-published CSV, pass@1 column only. Human row and models without an unambiguous existing main identity omitted. No model freshness restamp.

44.5
5

Configuration: Pass@1; three seeds 36/43/50; all 125 tasks; mean success rate, not pass^3

Version: TuringSpace-Bench arXiv 2611.25387 v1Harness: Owner public computer-use agents; 125 tasks × test seeds 36,43,50; 1920x1080 screenshots; <=50 model calls per instanceEvaluator: TuringSpace authorsObserved: Nov 20, 2026Confidence: highSource

Notes: Owner-published CSV, pass@1 column only. Human row and models without an unambiguous existing main identity omitted. No model freshness restamp.

41.6
6

Configuration: Pass@1; three seeds 36/43/50; all 125 tasks; mean success rate, not pass^3

Version: TuringSpace-Bench arXiv 2611.25387 v1Harness: Owner public computer-use agents; 125 tasks × test seeds 36,43,50; 1920x1080 screenshots; <=50 model calls per instanceEvaluator: TuringSpace authorsObserved: Nov 20, 2026Confidence: highSource

Notes: Owner-published CSV, pass@1 column only. Human row and models without an unambiguous existing main identity omitted. No model freshness restamp.

37.1
7

Configuration: Pass@1; three seeds 36/43/50; all 125 tasks; mean success rate, not pass^3

Version: TuringSpace-Bench arXiv 2611.25387 v1Harness: Owner public computer-use agents; 125 tasks × test seeds 36,43,50; 1920x1080 screenshots; <=50 model calls per instanceEvaluator: TuringSpace authorsObserved: Nov 20, 2026Confidence: highSource

Notes: Owner-published CSV, pass@1 column only. Human row and models without an unambiguous existing main identity omitted. No model freshness restamp.

32.0
8

Configuration: Pass@1; three seeds 36/43/50; all 125 tasks; mean success rate, not pass^3

Version: TuringSpace-Bench arXiv 2611.25387 v1Harness: Owner public computer-use agents; 125 tasks × test seeds 36,43,50; 1920x1080 screenshots; <=50 model calls per instanceEvaluator: TuringSpace authorsObserved: Nov 20, 2026Confidence: highSource

Notes: Owner-published CSV, pass@1 column only. Human row and models without an unambiguous existing main identity omitted. No model freshness restamp.

30.7
9

Configuration: Pass@1; three seeds 36/43/50; all 125 tasks; mean success rate, not pass^3

Version: TuringSpace-Bench arXiv 2611.25387 v1Harness: Owner public computer-use agents; 125 tasks × test seeds 36,43,50; 1920x1080 screenshots; <=50 model calls per instanceEvaluator: TuringSpace authorsObserved: Nov 20, 2026Confidence: highSource

Notes: Owner-published CSV, pass@1 column only. Human row and models without an unambiguous existing main identity omitted. No model freshness restamp.

30.1
10

Configuration: Pass@1; three seeds 36/43/50; all 125 tasks; mean success rate, not pass^3

Version: TuringSpace-Bench arXiv 2611.25387 v1Harness: Owner public computer-use agents; 125 tasks × test seeds 36,43,50; 1920x1080 screenshots; <=50 model calls per instanceEvaluator: TuringSpace authorsObserved: Nov 20, 2026Confidence: highSource

Notes: Owner-published CSV, pass@1 column only. Human row and models without an unambiguous existing main identity omitted. No model freshness restamp.

5.9
11

Configuration: Pass@1; three seeds 36/43/50; all 125 tasks; mean success rate, not pass^3

Version: TuringSpace-Bench arXiv 2611.25387 v1Harness: Owner public computer-use agents; 125 tasks × test seeds 36,43,50; 1920x1080 screenshots; <=50 model calls per instanceEvaluator: TuringSpace authorsObserved: Nov 20, 2026Confidence: highSource

Notes: Owner-published CSV, pass@1 column only. Human row and models without an unambiguous existing main identity omitted. No model freshness restamp.

2.4
12

Configuration: Pass@1; three seeds 36/43/50; all 125 tasks; mean success rate, not pass^3

Version: TuringSpace-Bench arXiv 2611.25387 v1Harness: Owner public computer-use agents; 125 tasks × test seeds 36,43,50; 1920x1080 screenshots; <=50 model calls per instanceEvaluator: TuringSpace authorsObserved: Nov 20, 2026Confidence: highSource

Notes: Owner-published CSV, pass@1 column only. Human row and models without an unambiguous existing main identity omitted. No model freshness restamp.

0.5

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

  • There is a fresh timestamped snapshot (or multiple snapshots) for this benchmark.
  • The model list covers the same version family you can actually deploy today.
  • Top candidates overlap with your required routing and feature requirements.

Be cautious when

  • There is only one benchmark snapshot or the dataset appears stale.
  • The benchmark metric direction is opposite of your decision objective.
  • The score difference between options is narrow and likely within implementation variance.

Related benchmarks

Last reviewed: Nov 20, 2026