TIR-Bench: leaderboard
A tool-calling and multimodal interaction benchmark for testing visual instruction following and execution reliability.
Metric: Score (imported). Source: arxiv.org. Source: benchmarklist.com. Saturation forecast: Around November 2026. 23 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 28.9 | #152 |
| 2 | O3 | 26.9 | #124 |
| 3 | Gemini 2.5 Flash | 25.2 | #245 |
| 4 | Grok 4 | 22.5 | #181 |
| 5 | InternVL3-78B | 21.4 | #352 |
| 6 | O4 Mini | 21.2 | #180 |
| 7 | InternVL3-38B | 19.1 | #405 |
| 8 | GPT-4.1 | 18.8 | #248 |
| 9 | GPT-4o | 17.3 | #343 |
| 10 | InternVL3-8B | 16.9 | #620 |
| 11 | Qwen 2.5 VL 7B | 16 | #643 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=tir-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.