— LLM Leaderboard —

Ranked by the people
who use them.

Community-driven rankings of the leading large language models, scored by accredited AI practitioners across the capabilities that matter in real work — reasoning, engineering and conversation. No vendor benchmarks, no marketing claims. Just the verdict of the people who build with them every day.

— Overall standings —

The models out
in front.

Anthropic

Claude Fable 5

98/100

Overall

Anthropic

Claude Opus 4.8

97/100

Overall

OpenAI

GPT-5.5

95/100

Overall

— Capability by capability —

Where each model
actually wins.

The top five performers in every category, scored on the work practitioners do most.

Professional Reasoning

Complex problem-solving and analytical judgement in real-world scenarios.

Claude Fable 598
Claude Opus 4.896
Gemini 3.1 Pro95
GPT-5.595
Claude Opus 4.794

Software Engineering

Code generation, debugging and algorithmic thinking under real constraints.

Claude Fable 599
Claude Opus 4.897
Claude Opus 4.794
GPT-5.594
Gemini 3.1 Pro92

Conversational Intelligence

Reading intent, tone and nuance across natural language exchanges.

Claude Opus 4.897
Claude Fable 596
Gemini 3.1 Pro96
GPT-5.595
Claude Opus 4.793

— Top ranked —

The top-rated
models.

The highest-rated models, side by side. Tap any score column to re-rank the board by that capability.

#	Model
1	Claude Fable 5 Anthropic	98	99	96	98
2	Claude Opus 4.8 Anthropic	96	97	97	97
3	GPT-5.5 OpenAI	95	94	95	95
4	Claude Opus 4.7 Anthropic	94	94	93	94
5	Gemini 3.1 Pro Google	95	92	96	94
6	Claude Opus 4.6 Anthropic	90	91	92	91
7	Gemini 3 Pro Google	92	89	93	91
8	GPT-5.4 OpenAI	92	90	92	91
9	GPT-5.2 OpenAI	90	88	91	90
10	Qwen3.7 Max Qwen	91	90	89	90
11	Claude Sonnet 4.6 Anthropic	89	88	91	89
12	DeepSeek V4 ProMax DeepSeek	90	89	88	89
13	GLM-5.1 Z.ai	89	89	88	89
14	Grok 4.20 xAI	90	88	90	89
15	Kimi K2.6 Moonshot AI	88	88	90	89
16	Gemini 3.5 Flash Google	88	88	89	88
17	GPT-5 OpenAI	88	85	90	88
18	GPT-5.1 OpenAI	89	84	89	87
19	Claude Sonnet 4.5 Anthropic	85	86	87	86
20	Gemini 3 Flash Google	86	84	88	86