Skip to main content

GPT-6.1 Sol security benchmark rundown

GPT-6.1 Sol (OpenAI) ranks 7th of 26 on BlueBench with 76.7 overall at $1.30 per task. Track scores, value against peers, strengths, weaknesses, and when to use it.

OpenAI · Closed weight · Tested Sep 30, 2026 – Oct 1, 2026 · Data updated Oct 2026
Methodology

Verdict

GPT-6.1 Sol ranks 7th of 26 on BlueBench with an overall score of 76.7, in the upper half of the leaderboard and 6.5 points behind Claude Opus 5.5. At $1.30 per task it is the 13th cheapest of 26, and Claude Sonnet 5.5 scores higher (77.5) for less ($1.01). Against the 8 models priced within 2x of its cost per task, it scores 9.0 points above the peer median overall; it is strongest against them in incident response. Malware Analysis, weighted 10%, is a secondary strength (+9.7 vs the peer median). Its results are among the most consistent across the benchmarks (reliability 52.9, 3rd of 26).

BlueBench score
76.7
#7 of 26 models
Cost per task
$1.30
#13 cheapest of 26
Reliability
52.9
#3 of 26 models
Overall vs cost peers
+9.0
Off the cost–score frontier
All GPT-6.1 Sol results and FAQ →

Capability by track

Incident response, threat hunting, and detection engineering each carry 30% of the BlueBench score; malware analysis carries 10%.

GPT-6.1 Sol scores 76.7 overall, 7th of 26 on the capability leaderboard. Its best rank is in malware analysis (87.5, tied 4th of 26) and its worst is in threat hunting (61.7, 13th of 26).

TrackWeightScoreRankPeer medianvs peers
Incident Response30%86.4#4 of 2665.6+20.8
Threat Hunting30%61.7#13 of 2655.0+6.7
Detection Engineering30%78.5#10 of 2676.7+1.8
Malware Analysis10%87.5#4 of 26 (tied)77.8+9.7

Cost, reliability, and value

Cost is USD per evaluated task including repeated attempts. Reliability is 100 minus the mean score range across the six benchmarks.

GPT-6.1 Sol averages $1.30 per task, including repeated attempts: the 13th cheapest of 26 models with a cost figure. It is off the cost–score frontier: Claude Sonnet 5.5 scores 77.5 at $1.01. Its peer group is the 8 models priced within 2x of its cost per task. Their median overall score is 67.7, so GPT-6.1 Sol is 9.0 points ahead of its price range. Reliability, 100 minus its mean score range across the benchmarks, is 52.9 (3rd of 26).

Strengths and weaknesses

Each track compared with the median of the 8 models priced within 2x of its cost per task. A gap of 5 points or more counts.

Against the 8 models priced within 2x of its cost per task, GPT-6.1 Sol is strongest in incident response (+20.8), threat hunting (+6.7) and malware analysis (+9.7). Differences are versus the peer median score in each track; ±5 points is the threshold.

Beats its cost peers

  • Incident Response86.4 (+20.8)
  • Threat Hunting61.7 (+6.7)
  • Malware Analysis87.5 (+9.7)

Lags its cost peers

No track by 5+ points

Nearest neighbors on the leaderboard

The models ranked just above and below it overall, with the largest track gaps.

  • Claude Sonnet 5.5 sits 2 places above, 0.8 points higher overall (77.5), at 78% of the cost per task ($1.01); it is 5.2 points ahead in threat hunting.
  • GPT-5.6 Sol sits one place above, 0.5 points higher overall (77.2), at 2.2x the cost per task ($2.89); no track differs by 5 points or more.
  • Claude Opus 4.8 sits one place below, 1.8 points lower overall (74.9), at 1.2x the cost per task ($1.56); it is 6.9 points ahead in threat hunting and 18.0 points behind in incident response.
  • DeepSeek V4.1 Flash sits 2 places below, 5.2 points lower overall (71.5), at 13% of the cost per task ($0.17); it is 6.8 points ahead in threat hunting and 25.5 points behind in incident response.
RankModelScore$/task
5Claude Sonnet 5.577.5$1.01
6GPT-5.6 Sol77.2$2.89
7GPT-6.1 Sol76.7$1.30
8Claude Opus 4.874.9$1.56
9DeepSeek V4.1 Flash71.5$0.17

Best for

Alternatives are limited to models costing at most 1.5x as much, except the overall leader and the cheaper-model pick.

  • Use itUse it for incident response (86.4, +20.8), threat hunting (61.7, +6.7) and malware analysis (87.5, +9.7) at its price: those tracks beat the median of the 8 models priced within 2x of its cost per task by 5 points or more.
  • InsteadUse Claude Sonnet 5.5 instead in most cases: it scores 77.5 overall (+0.8) at $1.01 per task (78% of the cost per task).
  • InsteadUse Claude Opus 4.8 instead if threat hunting is the priority: it scores 68.6 there (vs 61.7) at $1.56 per task.
  • InsteadUse Claude Opus 5.5 instead if the highest overall score matters more than cost: it scores 83.2 (6.5 points higher) at $3.54 per task.

Notable benchmark results

Each BlueBench benchmark is a separate intrusion, real or simulated, with its own tasks and grading.

  • 3rd place on Intrusion-002 (Windows Enterprise Intrusion): 79.6 (3rd of 26).
  • Its scores range from 52.0 on Simulation-001 to 93.0 on Intrusion-001, a spread of 41.0 points.
BenchmarkScoreRank
Simulation-003Impact & Exfiltration81.2#5 of 26
Simulation-002Identity & Active Directory Attacks77.1#7 of 26
Simulation-001Initial Access & Command-and-Control52.0#13 of 26
Intrusion-003AWS Cloud Intrusion85.6#5 of 26
Intrusion-002Windows Enterprise Intrusion79.6#3 of 26
Intrusion-001macOS Threat Investigation93.0#5 of 26

About this rundown

Provider
OpenAI · closed weight
Tested
Sep 30, 2026 – Oct 1, 2026
Suite
74 tasks across 6 benchmarks, evaluated in a common harness
Data updated
Oct 27, 2026
How it is written
Generated from the BlueBench result files with fixed rules: cost peers are models priced within 2x per task (or the nearest by cost when fewer than three are), a track is a strength or weakness at 5+ points from the peer median, and ranks share a place on ties. The text is regenerated whenever the results change.

AI for the blue team.

Run Cotool's harness in your environment to get real security work done

Book a demo