Ko-AgentBench - L6 Efficient Tool Utilization — leaderboard
Metric: Efficiency Score (%). Source: huggingface.co. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Flash | 33.33 |
| 2 | DeepSeek V3.1 | 16.67 |
| 3 | GPT-5 | 14.44 |
| 4 | Gemini 2.5 Pro | 13.33 |
| 5 | Gemini 2.5 Flash Lite | 13.33 |
| 6 | GPT-5 Mini | 6.67 |
| 7 | Grok 4.1 Fast | 6.67 |
| 8 | GPT-4o | 0 |
| 9 | GPT-4o Mini | 0 |
| 10 | Claude Sonnet 4.5 | 0 |
| 11 | Claude Haiku 4.5 | 0 |
| 12 | nova-2-lite-v1 | 0 |
| 13 | Qwen 3 Next 80B A3B | 0 |
Interactive version: aibenchmarks.dev/benchmark?slug=ko-agentbench-l6-efficient-tool-utilization · How the rankings work · Data refreshed daily, snapshot 2026-07-17.