AI BENCHY
Advertise here
#115

DeepSeek V3.2

DeepSeek Release: 2025-12-01 Tested on: 2026-05-22 00:35 deepseek/deepseek-v3.2::none
671B total (37B active)MoEOpen source
(medium) (none)

Summary

DeepSeek V3.2 scores 5.6 on AI BENCHY and ranks #115. It has 10.0 reliability, a 46.7% pass rate, $0.018 total cost, and 14.46s average response time.

What makes DeepSeek V3.2 unique: It stands out most in General Intelligence, where it ranks #4, while Anti-AI Tricks is its weakest area at #18. Its total benchmark cost is unusually low for its score range.

Model facts

Researched on 2026-08-12

Reported
Parameters
671B total (37B active)
Architecture
MoE
Availability
Open source
License
MIT
Parameter source Availability source

Score

5.6

Consistency

8.0

Reliability

10.0

Total Cost (Current Price)

$0.018

Total Output Tokens

11,163

Total Input Tokens

0

Input Price

$0.252 / 1M

Output Price

$0.378 / 1M

Tests Correct

Wrong Tests: 13

Attempt pass rate: 46.7%

Flaky tests

5

Flaky tests had mixed outcomes across runs (at least one pass and one fail).

Response Time (avg)

14.46s

Response Time (max): 115.89s

Response Time (total): 289.21s

Wrong answer: 8 Extra formatting: 2 Did not follow instructions: 2 Invalid tool call: 1

Hamster playing table tennis

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

#115 DeepSeek V3.2

none
Cost
$0.002
Time
7.0s
Tokens
1,046 tok

Run history

Tested on Score Reliability Tests Correct Total Cost Compare
2026-08-14 02:58 Re-test 5.0 10.0 $0.054 ↑ Compare
2026-07-16 23:15 New test added 5.0 10.0 $0.054 ↑ Compare
2026-06-04 14:22 New test added 5.2 10.0 $0.017 ↓ Compare
2026-05-22 00:35 Suite changed 5.6 10.0 $0.018 Current run
2026-05-08 15:31 Suite changed 5.7 10.0 $0.016 Compare
2026-04-20 17:48 First recorded run 6.1 N/A $0.016 Compare

This run used a different benchmark suite. Keep suite changes in mind when reading historical movement.

Charts

Choose the first model, then click a second model to open a side-by-side page.

Total Output Tokens

Score vs Total Output Tokens

Quick Compare

DeepSeek V3.2nonevsLaguna S 2.1lowFree AvailableDeepSeek V3.2nonevsGPT-4o-mininoneDeepSeek V3.2nonevsMiniMax M2.7mediumDeepSeek V3.2nonevsLing-2.6-flashnoneDeepSeek V3.2nonevsNorth Mini CodenoneFree AvailableDeepSeek V3.2nonevsNemotron 3 SupernoneFree AvailableDeepSeek V3.2nonevsQwen3.5-9BnoneDeepSeek V3.2nonevsHy4 previewnoneDeepSeek V3.2nonevsMiMo-V2.5noneDeepSeek V3.2nonevsRing-2.6-1Tnone

Category Breakdown

Category Score Consistency Tests Correct
Anti-AI Tricks 3.3 8.2
Coding 3.1 5.4
Combined 6.5 10.0
Data parsing and extraction 6.3 5.8
Domain specific 3.0 6.9
General Intelligence 7.6 10.0
Instructions following 10.0 10.0
Puzzle Solving 7.5 7.7
Tool Calling 10.0 10.0
Trivia 3.0 10.0

Compared models