Gemma 4 31B vs Kimi K2 Thinking

Change models
Kimi K2 ThinkingMoonshot AI
IndexAll categories, out of 100.
47.6 ±3.3
47.8 ±8.6 best
GapIndex points below the top.
−0.2 · no clear gap
Top
CoverageIndex weight that has results.
100% best
65%
SpeedOutput tokens per second.
36/s
40/s best
EloLMArena rating and rank.
1442
—
ReleasedMonth of release.
2026-04
2025-11
Input / 1MUS dollars per 1M input tokens.
Free best
$0.60
Output / 1MUS dollars per 1M output tokens.
Free best
$2.50
1K chats2K tokens in, 500 out, each.
Free best
$2.45
1K documents20K tokens in, 1K out, each.
Free best
$14.50
1K agent tasks200K tokens in, 20K out, each.
Free best
$170.00
ContextMaximum tokens in one request.
256K
262K best
AgenticMulti-step tasks with tools.
44.8
47.8 best
CodingCode writing and repair.
49.5 best
44.5
ReasoningLogic problems and puzzles.
46.2
—
MultimodalTasks with images and text.
56.3
—
KnowledgeFacts and expert knowledge.
44.7
48.9 best
MultilingualTasks in many languages.
47.5
—
InstructionTasks with strict prompt rules.
58.7
—
MathMath problems.
51.0 best
50.0

Benchmarks

5 shared
Only Gemma 4 31B: 42Only Kimi K2 Thinking: 17
Gemma 4 31B
Kimi K2 Thinking
75.8%106 / 220
84.2% best78 / 220
73.3%97 / 203
83.1% best83 / 203
39.3% best35 / 66
37.1%38 / 66
52.6%42 / 53
65.7% best10 / 53
50.8% best27 / 51
36.6%42 / 51