🏆 The AI Podium

The reference ranking of the best AI models, updated every month.

September 2026 edition

🏆 The AI Podium

Official ranking · September 2026

View methodology See full podium

Generalists

Best all-rounders

Claude Fable 5

Anthropic·93/100·$10/$50Premium

Claude Opus 5

Anthropic·93/100·$5/$25Premium

Muse Spark 1.1

Meta·93/100·$1.25/$4.25Standard

GPT-5.6 Sol+4

OpenAI·92/100·$4/$20Standard

Gemini 3.6 Flash-3

Google·91/100·$0.75/$3.75Low cost

Code Specialists

SWE-bench · Terminal-Bench · HumanEval

GPT-5.6 Sol

OpenAI·89/100·$4/$20Standard

Gemini 3.6 Flash+2

Google·84/100·$0.75/$3.75Low cost

Claude Fable 5+4

Anthropic·84/100·$10/$50Premium

GPT-5.5-1

OpenAI·80/100·$5/$30Premium

Claude Opus 4.8-3

Anthropic·79/100·$5/$25Premium

Vision Specialists

Multimodal & image

Gemini 3.1 Pro

Google·80/100·$2/$12Standard

Gemini 3 Pro+2

Google·76/100·$1.25/$5Standard

GPT-5.5+4

OpenAI·72/100·$5/$30Premium

Claude Opus 4.8-1

Anthropic·71/100·$5/$25Premium

Claude Opus 4.7

Anthropic·68/100·$5/$25Premium

Multilingual Specialists

FR · EN · CN & more

Qwen 3 Max

Alibaba·81/100·$0.78/$3.9Low cost

Mistral Large 3+2

Mistral AI·79/100·$2/$6Standardapache-2.0

Gemini 3.1 Pro

Google·68/100·$2/$12Standard

GPT-5.5+4

OpenAI·67/100·$5/$30Premium

GLM-5-3

Zhipu AI·67/100·$0.5/$2Low costapache-2.0

Best value for money

Generalists · capability vs cost

Gemini 3.6 Flash+4

Google·88/100·$0.75/$3.75Low cost

Qwen3.7 Plus-0

Alibaba·85/100·$0.4/$1.6Low cost

Muse Spark 1.1+3

Meta·83/100·$1.25/$4.25Standard

Kimi K2.6-1

Moonshot AI·82/100·$0.95/$4Low costmodified-mit

Claude Sonnet 5

Anthropic·80/100·$2/$10Standard

Best value for money · Code

Most coding power per dollar

Gemini 3.6 Flash

Google·84/100·$0.75/$3.75Low cost

Kimi K2.6

Moonshot AI·76/100·$0.95/$4Low costmodified-mit

DeepSeek V3.2-0

DeepSeek·76/100·$0.28/$0.42Low costmit

Qwen 3 Max+4

Alibaba·72/100·$0.78/$3.9Low cost

DeepSeek V3

DeepSeek·72/100·$0.27/$1.1Low costmit

Open Source

Self-hostable, free license

Kimi K2.6+1

Moonshot AI·72/100·$0.95/$4Low costmodified-mit

Mistral Small 3-0

Mistral AI·61/100·$0.2/$0.6Low costapache-2.0

Mistral Large 3

Mistral AI·60/100·$2/$6Standardapache-2.0

DeepSeek R2+4

DeepSeek·59/100·$0.55/$2.19Low costmit

DeepSeek V3-3

DeepSeek·59/100·$0.27/$1.1Low costmit

Local · Laptop

Light · < 16 GB RAM

Gemma 4 12B

Google·89/100apache-2.0

Granite 4.1 8B

IBM·81/100apache-2.0

Qwen3 8B

Alibaba·80/100apache-2.0

Mistral Nemo 12B

Mistral AI·75/100apache-2.0

Ministral 3 8B

Mistral AI·74/100apache-2.0

Local · Desktop

~24 GB RAM

Qwen3.6 35B-A3B+1

Alibaba·92/100apache-2.0

Qwen3.6 27B+3

Alibaba·89/100apache-2.0

Gemma 4 31B-0

Google·89/100apache-2.0

Gemma 4 26B-A4B-2

Google·87/100apache-2.0

Qwen3 32B

Alibaba·85/100apache-2.0

Local · Workstation/Server

Big VRAM

GLM-5.2

Zhipu AI·91/100mit

Kimi K2.7 Code

Moonshot AI·88/100modified-mit

DeepSeek V4-Flash-1

DeepSeek·88/100mit

Qwen3.5 397B-A17B-2

Alibaba·82/100apache-2.0

DeepSeek V3

DeepSeek·82/100·$0.27/$1.1Low costmit

🔬 How is this ranking made?

A transparent algorithm combines official benchmarks, pricing, context window and freshness to score each model.

View full methodology →

📅 Monthly archives

September 2026 (current)