Gaia2

Gaia2 - general AI assistant benchmark measuring multi-step agentic task completion.

Category

General Agents

Max Score

100.0

Score Type

percent

Active

Yes

Model Scores

Scores for Gaia2

Model Score Date Verified
Muse-Glimmer-30B
27.01%
01.08.2026 Verified
Gemma4-31B-it
15.92%
01.08.2026 Unverified
Qwen3.6-27B
21.70%
01.08.2026 Unverified
MiniCPM5-2B
100.00%
— Verified
LFM2.5-2.6B
36.98%
— Verified
Qwen3.5-2B
34.41%
— Verified
Gemma4-E2B
5.79%
— Verified
Qwen3.5-4B
83.76%
— Verified
granite-4.2-3B
49.52%
— Verified
Nemotron-3-Nano-4B
26.50
— Verified
LFM2.5-8B-A1B
23.47%
— Verified
Gemma4-E4B
20.90%
— Verified