Chat2Workflow - Research: leaderboard
Metric: Resolve rate (%) on Chat2Workflow research tasks: share of test cases where the workflow the model generates for the current round of a multi-turn request, deployed on Dify 1.9.2, executes and its output satisfies every requirement of the instruction, judged by GPT-5.1; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 62.96 |
| 2 | GLM-4.6 | 57.04 |
| 3 | Claude Sonnet 4.5 | 56.3 |
| 4 | GLM-4.7 | 54.82 |
| 5 | GPT-5.2 | 53.33 |
| 6 | Kimi K2 (Thinking) | 44.44 |
| 7 | DeepSeek V3.1 | 42.22 |
| 8 | GPT-5.1 | 41.48 |
| 9 | DeepSeek V3.2 | 40.74 |
| 10 | Kimi K2 | 40.74 |
| 11 | Qwen 3 235B A22B | 32.59 |
| 12 | Qwen 3 Coder 480B A35B Instruct | 29.63 |
| 13 | Qwen 3 32B | 22.22 |
| 14 | Qwen 3 8B | 21.48 |
| 15 | Qwen 3 14B | 18.52 |
Interactive version: theaggregate.ai/benchmark?slug=chat2workflow-research · How It Works · Data refreshed daily, snapshot 2026-10-07.