TIDE Trajectory - BlocksWorld - Success: leaderboard
Metric: Success Rate (%). Source: arxiv.org. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 100 |
| 2 | GPT-OSS-120B | 100 |
| 3 | DeepSeek R1 | 100 |
| 4 | Gemini 2.5 Flash (Non-reasoning) | 99 |
| 5 | DeepSeek V3.2 (Non-reasoning) | 98 |
| 6 | Llama 3.3 70B Instruct | 96 |
| 7 | GLM-4 32B (0414) | 83 |
| 8 | Phi-4 (Reasoning) | 68 |
| 9 | Phi-4 | 56 |
| 10 | Llama 3.1 8B Instruct | 32 |
| 11 | GLM-4 9B Chat | 23 |
Interactive version: theaggregate.ai/benchmark?slug=tide-trajectory-blocksworld-success · How It Works · Data refreshed daily, snapshot 2026-09-19.