Navigate
AI BENCHY
Advertise here

Gemini 3.1 Pro Preview (medium) vs GPT-6 Astra (xhigh)

GPT-6 Astra (xhigh) leads on average score with 9.9 vs 9.2. Gemini 3.1 Pro Preview (medium) has the lower benchmark cost at $1.352 vs $5.156. Gemini 3.1 Pro Preview (medium) is faster at 20.62s vs 28.44s, with pass rates of 90.9% vs 97.0%.

Last updated at: 2026-09-04

Rank
#18
Total Output Tokens
97,238
Response Time (avg)
20.62s
Total Cost
$1.352
Rank
#5
Total Output Tokens
87,183
Response Time (avg)
28.44s
Total Cost
$5.156
Recommended model Gemini 3.1 Pro Preview (medium)

Its score stays close to the best score here (9.2 vs 9.9), while costing about 3.8x less than GPT-6 Astra (xhigh).

Detailed comparison

Metric Gemini 3.1 Pro Preview Gemini 3.1 Pro Preview medium Release: 2026-02-19 GPT-6 Astra GPT-6 Astra xhigh Release: 2026-09-04
Score 9.2 9.9
Rank #18 #5
Reliability 10.0 9.9
Consistency 10.0 9.6
Attempts 66/66 66/66
Tests Correct
Attempt pass rate 90.9% 97.0%
Flaky tests 0 1
Total Runs 66 66
Cost per result 6.758 24.552
Total Cost $1.352 $5.156
Input Price $2.000 / 1M $10.000 / 1M
Output Price $12.000 / 1M $50.000 / 1M
Total Input Tokens 92,296 79,662
Output Tokens 5,232 5,055
Reasoning Tokens 92,006 82,128
Response Time (avg) 20.62s 28.44s
Response Time (max) 88.68s 381.11s
Response Time (total) 329.94s 625.68s
Parameters ~1.2T total (~20B active) ~2T total (~150B active)
Availability Closed Closed

Model generation showcase

Hamster playing table tennis

Prompt: Create a detailed SVG illustration of a hamster playing table tennis.

#18 Gemini 3.1 Pro Preview

medium
Cost
$0.115
Time
87.2s
Tokens
9,629 tok

#5 GPT-6 Astra

xhigh
Cost
$0.694
Time
257.5s
Tokens
13,964 tok

Top Models by Score

Score vs Total Cost

Response Time (avg)

Score vs Response Time (avg)

Total Output Tokens

Score vs Total Output Tokens

Category Breakdown

Coding Score Consistency Attempt pass rate Flaky tests Tests Correct Response Time (avg) Input Tokens Output Tokens Reasoning Tokens
Gemini 3.1 Pro Preview 7.9 9.9 66.7% 0 40.17s 8,124 435 41,247
GPT-6 Astra 10.0 10.0 100.0% 0 8.27s 7,302 396 4,031

Quick Compare

Switch Comparison Pair