Skip to main content
AITier
Home
Rankings
Models
Providers
Widgets
Test
EN
Search models / rankings / scenarios
AI Model Rankings
Browse and compare AI models with detailed rankings across multiple domains
Back to Home
General Ranking
Code Ranking
Math Reasoning
Science
Reasoning
Agentic Capability
Multimodal Ranking
Image Generation
Text-to-Video
Image-to-Video
Audio
Rank
Model
Score
Rank
Model
Score
1
Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)
96.0
2
Claude Opus 5 (Adaptive Reasoning, Max Effort)
96.0
3
GPT-5.5 (medium)
96.0
4
GPT-5.6 Sol (low)
96.0
5
GPT-5.6 Terra (medium)
95.0
6
Claude Opus 4.8 (Adaptive Reasoning, Max Effort)
94.0
7
GPT-5.4 (xhigh)
94.0
8
Claude Opus 4.7 (Adaptive Reasoning, Max Effort)
93.0
9
Claude Sonnet 5 (Non-reasoning, High Effort)
93.0
10
GPT-5.2 (Non-reasoning)
93.0
11
Grok 4.6 (high)
93.0
12
DeepSeek V4 Pro 0813 (Reasoning, Max Effort)
Open
91.0
13
Gemini 3.1 Pro Preview
91.0
14
GPT-5.4 nano (Non-Reasoning)
91.0
15
Grok 4.5 (high)
91.0
16
Muse Spark 1.2 (xhigh)
91.0
17
Qwen3.8 Max
Open
91.0
18
GLM-5.2
Open
90.0
19
GPT-5.2 Codex (xhigh)
89.0
20
Gemini 3.5 Flash (high)
88.0
35 items
1
2
20 / page
Data source: LiveBench (Abacus.AI, NYU, NVIDIA, UMD, USC, Columbia) · Mathematics (category) (release 2026-06-25)
· Updated: 2026-08-13