Navigate
AI BENCHY
AD
Track all your projects in one dashboard. Get 📊stats, 🔥heatmaps and 👀recordings in one self-hosted dashboard.
uxwizz.com

AI BENCHY Compare

DeepSeek: DeepSeek V4 Flash vs MoonshotAI: Kimi K3

Summary

Kimi K3 (max) leads on average score with 8.5 vs 8.3. DeepSeek V4 Flash (high) has the lower benchmark cost at $0.029 vs $3.112. DeepSeek V4 Flash (high) is faster at 45.85s vs 122.48s, with pass rates of 74.6% vs 82.5%.

Recommended modelDeepSeek V4 Flash (high)Its score stays close to the best score here (8.3 vs 8.5), while costing about 109.4x less than Kimi K3 (max).

Last updated at: 2026-07-16

Metric DeepSeek V4 Flash DeepSeek V4 Flash high Release: 2026-04-24 Kimi K3 Kimi K3 max Release: 2026-07-16
Score 8.3 8.5
Rank #30 #21
Reliability 10.0 9.4
Consistency 8.5 9.2
Tests Correct
Attempt pass rate 74.6% 82.5%
Flaky tests 4 2
Total Runs 63 63
Cost per result 0.299 19.446
Total Cost $0.029 $3.112
Input Price $0.098 / 1M $3.000 / 1M
Output Price $0.196 / 1M $15.000 / 1M
Total Input Tokens 39,745 34,916
Output Tokens 10,310 2,910
Reasoning Tokens 123,501 197,529
Response Time (avg) 45.85s 122.48s
Response Time (max) 218.13s 766.58s
Response Time (total) 962.79s 2327.06s

Generation showcase

Hamster playing table tennis

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

#30 DeepSeek V4 Flash

high
Cost
$0.003
Time
93.1s
Tokens
7,926 tok

#21 MoonshotAI: Kimi K3

max
Cost
$0.281
Time
506.9s
Tokens
18,863 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
DeepSeek V4 Flash 7.8 10.0 66.7% 0 50.60s 7,279 395 34,862
Kimi K3 10.0 10.0 100.0% 0 325.79s 8,061 460 84,012

Quick Compare

Switch Comparison Pair