GTA-Workflow - Leaf Success Rate: leaderboard

Leaf checkpoint success on GTA-Workflow, the open-ended tier of GTA-2 (2026): 132 long-horizon productivity workflows from real user queries with deployed tools and multimodal inputs, run end to end in a ReAct agent framework and judged by GPT-5.2 against recursive checkpoint rubrics: the share of leaf checkpoints scoring above 7 of 10. Percent; higher is better.

Metric: Leaf success rate (%): share of leaf checkpoints scoring above 7 of 10 on GTA-Workflow, the open-ended long-horizon workflow tier of GTA-2 (132 real-user tasks with deployed tools and multimodal inputs), run end to end in the default Lagent ReAct framework; deliverables judged by GPT-5.2 against recursive checkpoint rubrics; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 13 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 2.5 Pro28.46#276
2Llama 4 Scout27.51#1818
3GPT-526.3#178
4Claude Sonnet 4.526.21#261
5Qwen 3 235B A22B26.04#628
6DeepSeek V3.225.61#419
7Kimi K225.35#443
8Grok 425.17#319
9Llama 3.1 70B Instruct3.37#1571
10Llama 3.1 8B Instruct1.47#4600
11Qwen 3 30B A3B1.3#1156
12Llama 3.2 3B Instruct0.78#7241
13Qwen 3 8B0.69#1873

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

Interactive version: theaggregate.ai/benchmark?slug=gta-workflow-leaf-success-rate · How It Works · Data refreshed daily, snapshot 2026-11-21.