effect right
Hello
Resources / AI Explorer

Which AI is worth trying next?

Compare leading AI models, see what the tests actually measure, and explore practical guides to using AI at work.

Compare AI models

6 models / 3 tests

Epoch AI independent evaluations. Higher scores are better within each test. Exact model variants and settings are retained.
Exact variant / provider

Anthropic

Google DeepMind

OpenAI

xAI

Meta AI

Not measuredNot measured

Alibaba

Last checked 8 Oct 2026, 10:41 UTCSources & disclaimer

Independent evaluations by Epoch AI · 333 model variants · 4 tests · Snapshot: 8 Oct 2026

Highlights show the highest mean in your selection. Small differences may be within measurement uncertainty. Tap a score for its source or understand the scores.

Scores are mean percentages, rounded to one decimal place. Highlights use unrounded values. Comparisons use Epoch's evaluation family; individual model settings differ.

Epoch AI, Capabilities & benchmarking. Scores selected and reformatted by HIKOMORE. No endorsement implied. Four scientific, factual and maths tests; provider announcements are outside this first dataset.

CC BY 4.0Methodology & limitationsReport a correction

HIKOMORE is part of the Claude Partner Network and an OpenAI Select Partner. Partnerships do not affect the results.

AI for work. A closer look.

Claude vs ChatGPT

A practical comparison for proposals, documents and team work.

Gemini vs ChatGPT

Research and files, inside and outside Google Workspace.

Claude vs Gemini

Document-heavy work, useful answers and easier hand-offs.

effect bottom mobile