Skip to main content

Gemini 3.1 Pro security benchmark rundown

Gemini 3.1 Pro (Google) ranks 25th of 26 on BlueBench with 45.3 overall at $4.63 per task. Track scores, value against peers, strengths, weaknesses, and when to use it.

Google · Closed weight · Tested Sep 11, 2026 · Data updated Oct 2026
Methodology

Verdict

Gemini 3.1 Pro ranks 25th of 26 on BlueBench with an overall score of 45.3, in the bottom quarter of the leaderboard and 37.9 points behind Claude Opus 5.5. At $4.63 per task it is the 24th cheapest of 26, and Claude Opus 5.5 scores higher (83.2) for less ($3.54). Against the 8 models priced within 2x of its cost per task, it scores 28.8 points below the peer median overall; it trails them most in incident response. Malware Analysis, weighted 10%, is a secondary weak spot (−20.5 vs the peer median). Its results vary more than most across the benchmarks (reliability 19.2, 23rd of 26).

BlueBench score
45.3
#25 of 26 models
Cost per task
$4.63
#24 cheapest of 26
Reliability
19.2
#23 of 26 models
Overall vs cost peers
−28.8
Off the cost–score frontier
All Gemini 3.1 Pro results and FAQ →

Capability by track

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

Gemini 3.1 Pro scores 45.3 overall, 25th of 26 on the capability leaderboard. Its best rank is in threat hunting (43.2, 23rd of 26) and its worst is in detection engineering (55.8, 26th of 26).

TrackWeightScoreRankPeer medianvs peers
Incident Response30%32.6#25 of 2679.0−46.4
Threat Hunting30%43.2#23 of 2666.3−23.0
Detection Engineering30%55.8#26 of 2680.9−25.1
Malware Analysis10%58.3#25 of 2678.8−20.5

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.

Gemini 3.1 Pro averages $4.63 per task, including repeated attempts: the 24th cheapest of 26 models with a cost figure. It is off the cost–score frontier: Claude Opus 5.5 scores 83.2 at $3.54, and 21 other models are also both cheaper and at least as good. Its peer group is the 8 models priced within 2x of its cost per task. Their median overall score is 74.1, so Gemini 3.1 Pro is 28.8 points behind its price range. Reliability, 100 minus its mean score range across the benchmarks, is 19.2 (23rd 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, Gemini 3.1 Pro lags in incident response (−46.4), detection engineering (−25.1), threat hunting (−23.0) and malware analysis (−20.5). Differences are versus the peer median score in each track; ±5 points is the threshold.

Beats its cost peers

No track by 5+ points

Lags its cost peers

  • Incident Response32.6 (−46.4)
  • Detection Engineering55.8 (−25.1)
  • Threat Hunting43.2 (−23.0)
  • Malware Analysis58.3 (−20.5)

Nearest neighbors on the leaderboard

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

  • DeepSeek V4 Pro sits 2 places above, 8.3 points higher overall (53.6), at 20% of the cost per task ($0.91); it is 19.0 points ahead in incident response.
  • Claude Opus 5 sits one place above, 5.2 points higher overall (50.5), at about the same cost per task ($4.44); it is 20.5 points ahead in detection engineering and 7.1 points behind in threat hunting.
  • Claude Sonnet 5 sits one place below, 0.5 points lower overall (44.8), at 38% of the cost per task ($1.77); it is 28.6 points ahead in malware analysis and 19.4 points behind in threat hunting.
RankModelScore$/task
23DeepSeek V4 Pro53.6$0.91
24Claude Opus 550.5$4.44
25Gemini 3.1 Pro45.3$4.63
26Claude Sonnet 544.8$1.77

Best for

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

  • Use itBlueBench does not single out a task for it: it is in the top 3 of no track and beats its cost peers by 5+ points in none.
  • InsteadUse Claude Opus 5.5 instead in most cases: it scores 83.2 overall (+37.9) at $3.54 per task (76% of the cost per task).

Notable benchmark results

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

  • Near the bottom on Simulation-003 (Impact & Exfiltration): 40.6 (26th of 26).
  • Near the bottom on Simulation-002 (Identity & Active Directory Attacks): 40.7 (25th of 26).
  • Near the bottom on Simulation-001 (Initial Access & Command-and-Control): 29.9 (25th of 26).
  • Near the bottom on Intrusion-003 (AWS Cloud Intrusion): 52.7 (24th of 26).
  • Near the bottom on Intrusion-002 (Windows Enterprise Intrusion): 47.7 (26th of 26).
  • Near the bottom on Intrusion-001 (macOS Threat Investigation): 63.7 (24th of 26).
  • Its scores range from 29.9 on Simulation-001 to 63.7 on Intrusion-001, a spread of 33.8 points.
BenchmarkScoreRank
Simulation-003Impact & Exfiltration40.6#26 of 26
Simulation-002Identity & Active Directory Attacks40.7#25 of 26
Simulation-001Initial Access & Command-and-Control29.9#25 of 26
Intrusion-003AWS Cloud Intrusion52.7#24 of 26
Intrusion-002Windows Enterprise Intrusion47.7#26 of 26
Intrusion-001macOS Threat Investigation63.7#24 of 26

About this rundown

Provider
Google · closed weight
Tested
Sep 11, 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