Kaggle ITBench: leaderboard
IBM's ITBench on Kaggle, which evaluates agents on IT-operations scenarios in site reliability engineering (SRE) and financial operations (FinOps). This board is Kaggle's Average Score, the mean of the FinOps and SRE scores, on a 0-100 scale where higher is better, kept only for models with both scores.
Metric: Score (%). Source: www.kaggle.com. Saturation forecast: Around June 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 58.33 | #71 |
| 2 | GPT-5.1 | 55.21 | #145 |
| 3 | Claude Opus 4.5 | 52.08 | #87 |
| 4 | GPT-5 | 51.88 | #105 |
| 5 | Gemini 2.5 Flash | 51.67 | #255 |
| 6 | O3 | 51.67 | #132 |
| 7 | Qwen 3 235B A22B 2507 Instruct | 46.04 | #315 |
| 8 | Claude Sonnet 4.5 | 45.83 | #148 |
| 9 | Gemini 2.5 Pro | 42.29 | #159 |
| 10 | Claude Haiku 4.5 | 42.08 | #287 |
| 11 | Grok 4.1 Fast (Reasoning) | 38.96 | #227 (Grok 4.1 Fast) |
| 12 | Mistral Small 3.1 | 32.92 | #623 |
| 13 | DeepSeek R1 0528 | 23.33 | #232 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=kaggle-itbench · How It Works · Data refreshed daily, snapshot 2026-11-05.