GAIA — Level 2 difficulty — GAIA public leaderboard, Level 2 — distinct tier from Level 1/3.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | GPT-5.2 | OpenAI | 74.8 | 1 | 2026-05-10 |
| 2 | DeepSeek-V3.2 | DeepSeek | 73.6 | 1 | 2026-05-10 |
| 3 | Gemini 3 Flash Preview | 66 | 1 | 2026-05-11 | |
| 4 | GPT-5 | OpenAI | 64.8 | 1 | 2026-05-10 |
| 5 | GPT-4o | OpenAI | 27 | 1 | 2026-05-10 |
| 6 | GPT-5 nano | OpenAI | 18.2 | 1 | 2026-05-10 |
| 7 | GPT-4 Turbo | OpenAI | 10.7 | 1 | 2026-05-11 |
| 8 | Gemma 4 26B A4B IT | 6.3 | 1 | 2026-05-16 | |
| 9 | GPT-4o mini | OpenAI | 4.4 | 1 | 2026-05-10 |
| 10 | Qwen2.5 7B | Alibaba | 0.6 | 1 | 2026-05-10 |
| 11 | GPT4 | OpenAI | 0 | 1 | 2026-05-10 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.