Benchmark
Gaia2
Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages. Higher is better.
- Models measured
- 13
- Scores recorded
- 18
- Sources
- 3
- IQ weight
- 1.07
Score distribution
- 23.4%
- Mean
- 21.9%
- Median
- 12.6%
- Std dev
- 7.4%
- 10th percentile
- 42.1%
- 90th percentile
- 4.4%
- Min
This benchmark has a weight of 1.07 for chat models, measured over 13 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Muse Glimmer 30B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 43.3% |
| 2 | Muse Glimmer 30Bhigh reasoning· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 43.3% |
| 3 | GPT 5high· Gaia2 (Meta ARE) data | 42.1% |
| 4 | Claude Sonnet 4thinking· Gaia2 (Meta ARE) data | 37.8% |
| 5 | Claude Sonnet 4· Gaia2 (Meta ARE) data | 34.8% |
| 6 | GPT 5low· Gaia2 (Meta ARE) data | 34.6% |
| 7 | Gemini 2.5 Pro· Gaia2 (Meta ARE) data | 25.8% |
| 8 | DeepSeek V3.1 Terminus· Gaia2 (Meta ARE) data | 23.1% |
| 9 | DeepSeek V3.1· Gaia2 (Meta ARE) data | 21.9% |
| 10 | Kimi K2· Gaia2 (Meta ARE) data | 20.1% |
| 11 | GPT 5minimal· Gaia2 (Meta ARE) data | 18.2% |
| 12 | Grok 4· Gaia2 (Meta ARE) data | 15.7% |
| 13 | Qwen3 235B (Web)thinking· Gaia2 (Meta ARE) data | 15.7% |
| 14 | GPT OSS 120Bhigh· Gaia2 (Meta ARE) data | 13.7% |
| 15 | Qwen3 235B (Web)· Gaia2 (Meta ARE) data | 11.6% |
| 16 | GPT 4o· Gaia2 (Meta ARE) data | 7.4% |
| 17 | Llama 4 Maverick· Gaia2 (Meta ARE) data | 7.4% |
| 18 | Llama 3.3 70B Instruct· Gaia2 (Meta ARE) data | 4.4% |
Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Data by LLM Stats