DeepSeek V3.1 vs Qwen3 235B-A22B

Change models
IndexAll categories, out of 100.
38.0 ±5.8 best
37.8 ±8.8
GapIndex points below the top.
Top
−0.2 · no clear gap
CoverageIndex weight that has results.
85% best
55%
SpeedOutput tokens per second.
98/s best
44/s
EloLMArena rating and rank.
1420 best
1366
ReleasedMonth of release.
2025-08
2025-04
Input / 1MUS dollars per 1M input tokens.
—
$0.455
Output / 1MUS dollars per 1M output tokens.
—
$1.82
1K chats2K tokens in, 500 out, each.
—
$1.82
1K documents20K tokens in, 1K out, each.
—
$10.92
1K agent tasks200K tokens in, 20K out, each.
—
$127.40
ContextMaximum tokens in one request.
128K
131K best
AgenticMulti-step tasks with tools.
38.3
—
CodingCode writing and repair.
42.7
43.7 best
ReasoningLogic problems and puzzles.
38.0
—
MultimodalTasks with images and text.
—
—
KnowledgeFacts and expert knowledge.
41.7 best
40.9
MultilingualTasks in many languages.
—
41.8
InstructionTasks with strict prompt rules.
24.9
30.9 best
MathMath problems.
29.9
31.8 best

Benchmarks

3 shared
Only DeepSeek V3.1: 14Only Qwen3 235B-A22B: 21
DeepSeek V3.1
Qwen3 235B-A22B
74.9% best82 / 144
62.9%108 / 144
49.8% best38 / 49
24.7%46 / 49
66.3% best21 / 28
40.1%24 / 28