Ko-AgentBench - L1 Single Tool Call — leaderboard
Metric: Argument Accuracy (%). Source: huggingface.co. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | nova-2-lite-v1 | 79.55 |
| 2 | Qwen 3 Next 80B A3B | 77.27 |
| 3 | Grok 4.1 Fast | 75 |
| 4 | GPT-5 | 72.73 |
| 5 | Gemini 2.5 Pro | 70.45 |
| 6 | GPT-5 Mini | 70.45 |
| 7 | GPT-4o Mini | 65.91 |
| 8 | Claude Sonnet 4.5 | 65.91 |
| 9 | Gemini 2.5 Flash Lite | 63.64 |
| 10 | Gemini 2.5 Flash | 61.36 |
| 11 | Claude Haiku 4.5 | 61.36 |
| 12 | GPT-4o | 59.09 |
| 13 | DeepSeek V3.1 | 54.55 |
Interactive version: aibenchmarks.dev/benchmark?slug=ko-agentbench-l1-single-tool-call · How the rankings work · Data refreshed daily, snapshot 2026-07-17.