TopBench (Agentic) - Filtering (F1): leaderboard

TopBench (2026) tests implicit predictive reasoning in table question answering: the answer is not in the table and must be predicted from historical rows. F1 from 0 to 1 of the record set the agent writes to a CSV file for filtering questions, solved by an agent that runs Python in a sandbox; higher is better.

Metric: F1 (0 to 1) on TopBench's filtering questions with classification targets, solved by the authors' ReAct agent that writes and runs Python in a Docker sandbox over the table: overlap between the set of records in the agent's CSV and the reference set of records whose predicted class meets the query; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Flash0.58#175
2Claude Sonnet 4.50.55#261
3DeepSeek V3.20.48#419
4Qwen 3 235B A22B 2507 (Thinking)0.42#466
5GPT-5.20.38#185
6Qwen 3 235B A22B 2507 Instruct0.28#616
7TableLLM-8B0
8TableLLM-13B0

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

Interactive version: theaggregate.ai/benchmark?slug=topbench-agentic-filtering-f1 · How It Works · Data refreshed daily, snapshot 2026-11-21.