Skip to main content

Gemini 3.6 Flash security benchmark rundown

Gemini 3.6 Flash (Google) ranks 19th of 26 on BlueBench with 62.7 overall at $3.01 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 3.6 Flash ranks 19th of 26 on BlueBench with an overall score of 62.7, in the lower half of the leaderboard and 20.5 points behind Claude Opus 5.5. At $3.01 per task it is the 19th cheapest of 26, and Claude Sonnet 5.5 scores higher (77.5) for less ($1.01). Against the 10 models priced within 2x of its cost per task, it scores 8.3 points below the peer median overall; it trails them most in incident response. Malware Analysis, weighted 10%, is a secondary weak spot (−19.5 vs the peer median). Its results vary more than most across the benchmarks (reliability 29.8, 19th of 26).

BlueBench score
62.7
#19 of 26 models
Cost per task
$3.01
#19 cheapest of 26
Reliability
29.8
#19 of 26 models
Overall vs cost peers
−8.3
Off the cost–score frontier
All Gemini 3.6 Flash 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 3.6 Flash scores 62.7 overall, 19th of 26 on the capability leaderboard. Its best rank is in detection engineering (81.1, 7th of 26) and its worst is in malware analysis (61.6, 23rd of 26).

TrackWeightScoreRankPeer medianvs peers
Incident Response30%54.5#21 of 2674.8−20.2
Threat Hunting30%53.0#18 of 2663.5−10.5
Detection Engineering30%81.1#7 of 2677.3+3.8
Malware Analysis10%61.6#23 of 2681.1−19.5

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 3.6 Flash averages $3.01 per task, including repeated attempts: the 19th 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, and 12 other models are also both cheaper and at least as good. Its peer group is the 10 models priced within 2x of its cost per task. Their median overall score is 71.0, so Gemini 3.6 Flash is 8.3 points behind its price range. Reliability, 100 minus its mean score range across the benchmarks, is 29.8 (19th of 26).

Strengths and weaknesses

Each track compared with the median of the 10 models priced within 2x of its cost per task. A gap of 5 points or more counts.

Against the 10 models priced within 2x of its cost per task, Gemini 3.6 Flash lags in incident response (−20.2), threat hunting (−10.5) and malware analysis (−19.5). 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 Response54.5 (−20.2)
  • Threat Hunting53.0 (−10.5)
  • Malware Analysis61.6 (−19.5)

Nearest neighbors on the leaderboard

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

  • GLM-5.2 sits 2 places above, 2.8 points higher overall (65.5), at 34% of the cost per task ($1.03); it is 20.3 points ahead in malware analysis and 5.2 points behind in detection engineering.
  • GPT-6 Luna sits one place above, 1.4 points higher overall (64.1), at 9% of the cost per task ($0.26); it is 16.6 points ahead in incident response and 11.2 points behind in detection engineering.
  • GPT-5.6 Luna sits one place below, 0.3 points lower overall (62.4), at 19% of the cost per task ($0.58); it is 17.5 points ahead in malware analysis and 13.6 points behind in detection engineering.
  • DeepSeek V4 Flash sits 2 places below, 4.1 points lower overall (58.6), at 6% of the cost per task ($0.17); it is 16.3 points behind in detection engineering.
RankModelScore$/task
17GLM-5.265.5$1.03
18GPT-6 Luna64.1$0.26
19Gemini 3.6 Flash62.7$3.01
20GPT-5.6 Luna62.4$0.58
21DeepSeek V4 Flash58.6$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 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 Sonnet 5.5 instead in most cases: it scores 77.5 overall (+14.8) at $1.01 per task (34% of the cost per task).
  • InsteadUse Claude Opus 5.5 instead if incident response is the priority: it scores 87.8 there (vs 54.5) at $3.54 per task.

Notable benchmark results

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

  • Its best standing is on Intrusion-001 (macOS Threat Investigation): 90.9 (11th of 26).
  • Its scores range from 46.3 on Simulation-001 to 90.9 on Intrusion-001, a spread of 44.5 points.
BenchmarkScoreRank
Simulation-003Impact & Exfiltration58.7#21 of 26
Simulation-002Identity & Active Directory Attacks65.5#14 of 26
Simulation-001Initial Access & Command-and-Control46.3#19 of 26
Intrusion-003AWS Cloud Intrusion68.7#18 of 26
Intrusion-002Windows Enterprise Intrusion60.2#22 of 26
Intrusion-001macOS Threat Investigation90.9#11 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