Inference
EN

Gemini 3.1 Pro vs GLM-5.3-Flash

Change models
Gemini 3.1 ProGoogle
GLM-5.3-FlashZ.AIBudget
IndexAll categories, out of 100.
62.3 ±3.1
63.1 ±2.2 best
GapIndex points below the top.
−0.8 · no clear gap
Top
CoverageIndex weight that has results.
95%
100% best
SpeedOutput tokens per second.
151/s best
122/s
EloLMArena rating and rank.
—
1471
ReleasedMonth of release.
2026-02
2026-08
Input / 1MUS dollars per 1M input tokens.
$2.00
$0.04 best
Output / 1MUS dollars per 1M output tokens.
$12.00
$0.078 best
1K chats2K tokens in, 500 out, each.
$10.00
$0.119 best
1K documents20K tokens in, 1K out, each.
$52.00
$0.876 best
1K agent tasks200K tokens in, 20K out, each.
$640.00
$9.55 best
ContextMaximum tokens in one request.
1M
1.05M best
AgenticMulti-step tasks with tools.
58.4
69.7 best
CodingCode writing and repair.
58.0
62.3 best
ReasoningLogic problems and puzzles.
65.1 best
58.9
MultimodalTasks with images and text.
63.5
65.0 best
KnowledgeFacts and expert knowledge.
69.4 best
57.7
MultilingualTasks in many languages.
—
64.0
InstructionTasks with strict prompt rules.
61.1 best
33.5
MathMath problems.
61.7 best
57.3

Benchmarks

9 shared
Only Gemini 3.1 Pro: 41Only GLM-5.3-Flash: 81
Gemini 3.1 Pro
GLM-5.3-Flash
GPQA diamond · Epoch AIKnowledge
94.4% best9 / 217
90.2%40 / 217
OTIS Mock AIME 2024-2025Math
95.6% best40 / 200
93.9%49 / 200
Chess PuzzlesReasoning
51.0% best18 / 90
21.0%59 / 90
FrontierMath-Tiers-1-3-v2-PrivateMath
59.6% best42 / 90
55.8%49 / 90
Mystery Game PuzzlesReasoning
34.0% best25 / 80
8.0%66 / 80
FrontierMath-Tier-4-v2-PrivateMath
26.8% best39 / 72
17.1%52 / 72
CharXiv ReasoningMultimodal
80.2%34 / 49
89.4% best11 / 49
MazeBench: visual spatial reasoning in a 3D open worldAgentic
1.0% best16 / 48
0.0%23 / 48
ParseBench: document parsing for AI agents (LlamaIndex leaderboard)Multimodal
67.3%17 / 39
70.5% best8 / 39