Gaia2
Gaia2 - general AI assistant benchmark measuring multi-step agentic task completion.
Category
General Agents
Score type
percent
Scale
100%
higher is better
Models tested
12
What is Gaia2?
Gaia2 is a benchmark for how well a model acts as a general-purpose agent. The official description says: "Gaia2 - general AI assistant benchmark measuring multi-step agentic task completion". Below we describe in simple words what benchmarks of this type generally examine.
What does it test?
Gaia2 measures how well a model acts as a general-purpose agent. In detail, it checks:
- solving long, multi-step tasks with many intermediate decisions
- using tools correctly (browsers, terminals, APIs, calendars, other apps)
- adapting when something unexpected happens mid-task
- staying on goal over long task horizons instead of wandering off
How to read the score
The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better.
Official methodology note: Gaia2 - general AI assistant benchmark measuring multi-step agentic task completion.
Frequently asked questions
What is the Gaia2 benchmark?
Gaia2 is a benchmark for how well a model acts as a general-purpose agent. The official description says: 'Gaia2 - general AI assistant benchmark measuring multi-step agentic task completion'. Below we describe in simple words what benchmarks of this type generally examine.
What does Gaia2 test?
Gaia2 mainly measures how well a model acts as a general-purpose agent. Concretely, it checks: solving long, multi-step tasks with many intermediate decisions; using tools correctly (browsers, terminals, APIs, calendars, other apps); adapting when something unexpected happens mid-task.
How is the Gaia2 score calculated?
The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better. The declared maximum score of this benchmark is 100.0.
Where do I get Gaia2 data and scores?
The official benchmark page/reference is linked on this page. Every score on Open LLM Radar is also available as open data: download the scores JSON (http://open-llm-radar-api:8000/api/v1/benchmarks/gaia2/scores) or pull it directly from the Open LLM Radar API with any HTTP client.
Benchmark data & downloads
Where to get, inspect and reuse the data behind this benchmark.
Benchmark dataset
The question/task data itself — download and evaluate locally
Benchmark metadata (JSON)
This page's benchmark record from the Open LLM Radar API
All model scores (JSON download)
Full score table as JSON; paginate with limit/offset params
Open LLM Radar API docs
Swagger UI to explore and reuse all radar API endpoints
Model scores 12 models tested
| # | Model | Score | Date | Source | Verified |
|---|---|---|---|---|---|
| 1 |
MiniCPM5-2B
OpenBMB
|
88.7%
|
— | Report | Verified |
| 2 |
Qwen3.5-4B
Qwen
|
78.6%
|
— | Report | Verified |
| 3 |
granite-4.2-3B
IBM
|
57.3%
|
— | Report | Verified |
| 4 |
LFM2.5-2.6B
LiquidAI
|
49.5%
|
— | Report | Verified |
| 5 |
Qwen3.5-2B
Qwen Team (Alibaba Cloud)
|
47.9%
|
— | Report | Verified |
| 6 |
Muse-Glimmer-30B
Meta Superintelligence Lab
|
43.3%
|
01.08.2026 | Meta Superintelligence Lab (official model card) | Verified |
| 7 |
LFM2.5-8B-A1B
LiquidAI
|
41.1%
|
— | Report | Verified |
| 8 |
Qwen3.6-27B
Qwen Team (Alibaba Cloud)
|
40%
|
01.08.2026 | Meta Superintelligence Lab (official model card) | Unverified |
| 9 |
Gemma4-E4B
Google
|
39.5%
|
— | Report | Verified |
| 10 |
Gemma4-31B-it
Google
|
36.4%
|
01.08.2026 | Meta Superintelligence Lab (official model card) | Unverified |
| 11 |
Gemma4-E2B
Google
|
30.1%
|
— | Report | Verified |
| 12 |
Nemotron-3-Nano-4B
NVIDIA
|
26.5%
|
— | Report | Verified |