TopBench - Treatment Effect: 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. Share of treatment-effect questions whose predicted direction of change matches the reference, from 0 to 1, answering from the serialized table without tools; higher is better.

Metric: Trend accuracy (0 to 1) on TopBench's 105 treatment-effect questions, which ask whether a counterfactual change to a profile raises or lowers its outcome given a historical table, answered from the serialized table (header kept, rows randomly sampled when it exceeds the context budget) without tools: share of questions whose predicted direction matches the reference; values and decisions extracted from the free-text answer by a DeepSeek-V3.2 judge and checked against the answer text; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 7 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Flash0.62#175
2DeepSeek V3.20.59#419
3Qwen 3 235B A22B 2507 Instruct0.55#616
4TableLLM-8B0.52
5GPT-5.20.51#185
6Qwen 3 235B A22B 2507 (Thinking)0.46#466
7TableLLM-13B0.35

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

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