Kimi K2.5 vs Ternary Bonsai 2 27B

Change models
Kimi K2.5Moonshot AI
IndexAll categories, out of 100.
52.2 ±2.2
52.2 ±5.0
GapIndex points below the top.
Top
Top
CoverageIndex weight that has results.
100% best
95%
SpeedOutput tokens per second.
36/s best
21/s
EloLMArena rating and rank.
—
—
ReleasedMonth of release.
2026-01
2026-09
Input / 1MUS dollars per 1M input tokens.
$0.532
$0.075 best
Output / 1MUS dollars per 1M output tokens.
$3.32
$0.50 best
1K chats2K tokens in, 500 out, each.
$2.73
$0.40 best
1K documents20K tokens in, 1K out, each.
$13.97
$2.00 best
1K agent tasks200K tokens in, 20K out, each.
$172.90
$25.00 best
ContextMaximum tokens in one request.
262K
262K
AgenticMulti-step tasks with tools.
49.4
54.0 best
CodingCode writing and repair.
53.9 best
52.2
ReasoningLogic problems and puzzles.
51.6
60.1 best
MultimodalTasks with images and text.
52.4 best
50.7
KnowledgeFacts and expert knowledge.
53.7 best
48.6
MultilingualTasks in many languages.
52.7
—
InstructionTasks with strict prompt rules.
53.5 best
52.6
MathMath problems.
53.8 best
51.1

Benchmarks

11 shared
Only Kimi K2.5: 78Only Ternary Bonsai 2 27B: 10
Kimi K2.5
Ternary Bonsai 2 27B
78.0% best70 / 217
77.0%82 / 217
95.9% best13 / 149
80.2%69 / 149
87.6% best41 / 141
85.8%50 / 141
76.8% best38 / 114
60.8%85 / 114
87.6% best47 / 84
85.8%56 / 84
93.9% best10 / 70
91.3%23 / 70
85.0%19 / 59
90.1% best9 / 59
96.1% best6 / 53
95.0%9 / 53
95.8%8 / 34
95.8%8 / 34
MMLU-ReduxKnowledge
94.5% best4 / 19
89.1%13 / 19
RealWorldQAMultimodal
81.0% best11 / 18
80.1%12 / 18