AI Labs
Which AI lab makes the best models? It depends on the metric. The lab with the single highest-ranked model is OpenAI (its OpenAI: GPT-5.5 leads the meo benchmark) — but other labs lead on Effective Value, cost-efficiency, or open-weights breadth. This page ranks every lab by its best meo rank, top model, and best 𝕍.
Every AI lab in the leaderboard — OpenAI, Anthropic, Google, DeepSeek, Qwen, xAI and more — with how many models they ship, their best meo rank, top model, and best Effective Value (𝕍).
Lab & model-maker concepts
Frequently asked questions
Which AI lab makes the best models?
It depends on the metric. One lab usually holds the single highest-accuracy model on the meo benchmark, but others lead on Effective Value, cost-efficiency, or the number of strong open-weights models. This page ranks every lab by its best meo rank, top model, and best 𝕍.
What is an AI lab?
An AI lab (or model maker) is the organization that trains and releases a model family — for example OpenAI, Anthropic, Google, Meta, DeepSeek, Qwen, Mistral, and xAI. Meo groups every leaderboard model by its creating lab so you can compare makers, not just individual models.
Which labs release open-weights models?
Several labs publish open-weights models you can self-host — including Meta (Llama), DeepSeek, Qwen/Alibaba, Mistral and others — while OpenAI, Anthropic and Google are primarily closed. The Weights column and the open-weights ranking show which is which.
How does meo rank AI labs?
Meo rolls up each lab’s models: how many it ships, its best meo rank, its top model, and its best Effective Value (𝕍). A lab can rank highly on a single flagship or on consistent quality across many models — both are shown.