Short answer: DeepSeek V4.1 Flash is 3.7x cheaper (0.099¢ vs 0.36¢ a chat). Qwen 3.8 Max ranks higher overall (#3 vs #8). Qwen 3.8 Max wins at reasoning, coding, maths, writing, agentic tasks. DeepSeek V4.1 Flash wins at following instructions. Pick DeepSeek V4.1 Flash for everyday use and Qwen 3.8 Max when quality matters most.
QQWENGANG
QWEN 3.8 MAX
QWEN GANG
CRAPGPT.LOLINDEPENDENT AI MODEL REVIEWS
CRAP SCORE733 SOURCES
BUILDAGENTSAUTOMATEIDEAS
CODEAGENTSIDEASINTO REALITY
POWER85ARENA · AA · CRAPGPT
Agentic
97
Reasoning
89
Writing
84
Math
83
Code
82
Value
50
BOSS★★★★
GENERALIST / AGENT
0.36¢ per chat · 275 chats per NZ$1
“Polished. Dangerous. Usually right.”
Every model is somebody.
— CRAPGPT.LOL —
DDEEPSEEKGANG
DEEPSEEK V4.1 FLASH
DEEPSEEK GANG
CRAPGPT.LOLINDEPENDENT AI MODEL REVIEWS
CRAP SCORE704 SOURCES
REALTIMEINTELLIGENCEFOR AFASTERTOMORROW
SEEPLANACTFASTER
POWER72ARENA · EPOCH · AA · CRAPGPT
Reasoning
85
Instruction
82
Code
73
Math
66
Writing
61
Value
66
OPERATOR★★★★
GENERALIST / REASONER
0.099¢ per chat · 1,009 chats per NZ$1
“Polished. Dangerous. Usually right.”
Every model is somebody.
— CRAPGPT.LOL —
Qwen 3.8 Max
DeepSeek V4.1 Flash
Cost per chat (NZD)
0.36¢
0.099¢
Chats per NZ$1
275
1,009
Overall rank
#3
#8
Vendor
Alibaba
DeepSeek
Speciality
General
General
Crowd Elo
988
988
Reasoning
89
85
Code
82
73
Math
83
66
Writing
84
61
Instruction
80
82
Agentic
97
41
Scores are 0 to 100 percentiles blending LMArena, Epoch AI, Artificial Analysis and CrapGPT's own tests. Updated 2026-10-11.
FAQ
Which is cheaper, Qwen 3.8 Max or DeepSeek V4.1 Flash?
DeepSeek V4.1 Flash costs 0.099¢ per typical chat versus 0.36¢ for Qwen 3.8 Max on CrapGPT (NZD, measured token use).
Which is better, Qwen 3.8 Max or DeepSeek V4.1 Flash?
Qwen 3.8 Max ranks higher overall on CrapGPT's blended benchmark score. By category: Qwen 3.8 Max leads at reasoning, coding, maths, writing, agentic tasks; DeepSeek V4.1 Flash leads at following instructions.
Can I try Qwen 3.8 Max and DeepSeek V4.1 Flash side by side?
Yes. CrapGPT's card battle sends one prompt to two models and lets you pick the winner, paying only for the two replies.
