Hy3 Preview vs MAI-Thinking-1

Change models
Hy3 PreviewTencent
IndexAll categories, out of 100.
52.7 ±7.4
52.7 ±8.1
GapIndex points below the top.
Top
Top
CoverageIndex weight that has results.
75% best
70%
SpeedOutput tokens per second.
81/s
—
EloLMArena rating and rank.
1401
—
ReleasedMonth of release.
2026-04
2026-06
Input / 1MUS dollars per 1M input tokens.
$0.18
—
Output / 1MUS dollars per 1M output tokens.
$0.60
—
1K chats2K tokens in, 500 out, each.
$0.66
—
1K documents20K tokens in, 1K out, each.
$4.20
—
1K agent tasks200K tokens in, 20K out, each.
$48.00
—
ContextMaximum tokens in one request.
262K best
256K
AgenticMulti-step tasks with tools.
55.3 best
45.3
CodingCode writing and repair.
54.6 best
54.4
ReasoningLogic problems and puzzles.
49.1
—
MultimodalTasks with images and text.
—
—
KnowledgeFacts and expert knowledge.
53.8 best
52.6
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
47.5
67.1 best
MathMath problems.
—
51.9

Benchmarks

5 shared
Only Hy3 Preview: 10Only MAI-Thinking-1: 7
Hy3 Preview
MAI-Thinking-1
87.2% best44 / 144
84.2%57 / 144
74.4% best48 / 114
73.5%52 / 114
87.2% best50 / 87
84.2%63 / 87
63.1%43 / 57
85.0% best2 / 57
54.4% best25 / 40
46.0%33 / 40