DeepSeek-R1-0528 vs DeepSeek V3.1

Change models
IndexAll categories, out of 100.
44.5 ±8.9 best
38.0 ±5.8
GapIndex points below the top.
Top
−6.5 · no clear gap
CoverageIndex weight that has results.
60%
85% best
SpeedOutput tokens per second.
34/s
98/s best
EloLMArena rating and rank.
1428 best
1420
ReleasedMonth of release.
2025-05
2025-08
Input / 1MUS dollars per 1M input tokens.
$0.571
—
Output / 1MUS dollars per 1M output tokens.
$2.29
—
1K chats2K tokens in, 500 out, each.
$2.29
—
1K documents20K tokens in, 1K out, each.
$13.71
—
1K agent tasks200K tokens in, 20K out, each.
$159.92
—
ContextMaximum tokens in one request.
164K best
128K
AgenticMulti-step tasks with tools.
—
38.3
CodingCode writing and repair.
46.5 best
42.7
ReasoningLogic problems and puzzles.
38.3 best
38.0
MultimodalTasks with images and text.
—
—
KnowledgeFacts and expert knowledge.
45.7 best
41.7
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
—
24.9
MathMath problems.
46.2 best
29.9

Benchmarks

6 shared
Only DeepSeek-R1-0528: 7Only DeepSeek V3.1: 11
DeepSeek-R1-0528
DeepSeek V3.1
81.0% best65 / 144
74.9%82 / 144
57.6%89 / 114
66.0% best79 / 114
85.0% best19 / 73
83.7%25 / 73
87.5% best18 / 49
49.8%38 / 49
91.4% best5 / 28
66.3%21 / 28
73.3% best3 / 14
56.4%9 / 14