Chat2Workflow - Research: leaderboard

Chat2Workflow (2026) tests whether language models build executable Dify workflows from multi-turn natural-language requests drawn from real business workflows, run against test cases. Resolve rate in percent on the research tasks; higher is better.

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 February 2027. 15 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Pro (Preview)62.96#119
2GLM-4.657.04#498
3Claude Sonnet 4.556.3#261
4GLM-4.754.82#380
5GPT-5.253.33#185
6Kimi K2 (Thinking)44.44
7DeepSeek V3.142.22#511
8GPT-5.141.48#240
9DeepSeek V3.240.74#419
10Kimi K240.74#443
11Qwen 3 235B A22B32.59#628
12Qwen 3 Coder 480B A35B Instruct29.63#652
13Qwen 3 32B22.22#972
14Qwen 3 8B21.48#1873
15Qwen 3 14B18.52#1269

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=chat2workflow-research · How It Works · Data refreshed daily, snapshot 2026-11-21.