AI Leaderboard Logo

AI Leaderboard

Compare the latest AI models across intelligence, benchmarks, speed and cost.

Top AI Models

RankModel Intelligence
#1
Gemini 4 Argon
Reasoning
58
#2
Claude Opus 5.5 (max)
Reasoning
56
#3
Muse Wonder (Full)
Reasoning
56
#4
Mistral Large 4 (Base (Preview))
Open Weights
54
#5
GPT-6 Astra (max)
Reasoning
53
#6
GPT-6 Astra (xhigh)
Reasoning
53
#7
Claude Fable 5.1 (max with fallback)
Reasoning
53
#8
Claude Fable 5.1 (xhigh with fallback)
Reasoning
53
#9
Muse Wonder (Standard (API))
Reasoning
53
#10
Claude Sonnet 5.5 (max)
Reasoning
52
#11
GPT-6 Astra (high)
Reasoning
51
#12
Claude Fable 5.1 (high with fallback)
Reasoning
51
#13
GPT-6.1 Sol (max)
Reasoning
51
#14
Claude Fable 5 (with fallback)
Reasoning
50
#15
MiMo-V2.6-Pro (max)
Open WeightsReasoning
50
#16
Claude Haiku 5.5
49
#17
Muse Spark 1.3 (max)
Open WeightsReasoning
48.1
#18
GPT-6 Instant
48
#19
GPT-5.6 Sol (max)
Reasoning
47
#20
Gemini 2.5 Flash-Lite (high)
45
#21
GLM-5.3 (max)
Open WeightsReasoning
45
#22
Grok 4 (4.6 high)
44.4
#23
Kimi K3 (max)
Reasoning
43.7
#24
GLM-5.3 (Flash)
Open Weights
42
#25
Qwen3.8 (2.4T A95B)
Open Weights
40
#26
Grok 4 (4.20 0309)
40
#27
Kolibri 78B MoE (78B)
Open Weights
38
#28
DeepSeek V4 Pro (0813)
Open Weights
36
#29
K2 Horizon 375B A23B (375B A23B)
Open Weights
34
#30
Mercury 2.5
30
#31
Llama 4 Scout
Open Weights
30
#32
Celeris-1
25
#33
Granite 4 (4.2 3B)
Open Weights
20
#34
EmbeddingGemma 2
Open Weights
20
#35
MiniCPM5-2B (2B)
Open Weights
14

Latest Models (September 2026)

GPT-6 Astra
Provider: openai
Intelligence:53
Context:1M
Claude Fable 5.1
Provider: anthropic
Intelligence:53
Context:1M
Claude Opus 5.5
Provider: anthropic
Intelligence:56
Context:1M
Gemini 2.5 Flash-Lite
Provider: google
Intelligence:45
Context:1M
Muse Spark 1.3Open
Provider: meta
Intelligence:48.1
Context:1M
K2 Horizon 375B A23BOpen
Provider: mbzuai
Intelligence:34
Context:524k
MiniCPM5-2BOpen
Provider: openbmb
Intelligence:14
Context:131k
MiMo-V2.6-ProOpen
Provider: xiaomi
Intelligence:50
Context:1M
Gemini 4 Argon
Provider: google
Intelligence:58
Context:1M
Claude Sonnet 5.5
Provider: anthropic
Intelligence:52
Context:1M
GPT-6.1 Sol
Provider: openai
Intelligence:51
Context:1M

Trending Research Papers

View all →

Dynamic Network Model from Partial Observations

Elahe Ghalebi, Baharan Mirzasoleiman, Radu Grosu et al. · 2018-05-27

• model• Open-Ended Question Answering

PAC-Bayes bounds for stable algorithms with instance-dependent priors

Omar Rivasplata, Emilio Parrado-Hernandez, John Shawe-Taylor et al. · 2018-06-18

• Deep Learning• Research

Automated Bridge Component Recognition using Video Data

Yasutaka Narazaki, Vedhus Hoskere, Tu A. Hoang et al. · 2018-06-18

• BIG-bench Machine Learning

Browse State-of-the-Art (SOTA)

View all →
Computer Vision
5 tasks
Natural Language Processing
4 tasks
Audio & Speech
3 tasks
Biology & Healthcare
2 tasks

Methodology

The AI Leaderboard ranks models using the AI Leaderboard Intelligence Index v1.0. The composite index incorporates 10 different evaluations, weighting them according to their predictive power and robustness against overfitting.

  • Agents — 30%
  • Coding — 20%
  • General — 30%
  • Scientific Reasoning — 20%