SciIR-Bench - Text: leaderboard

Metric: Accuracy (%) on the Text track (spelling and position of required text); a gemini-3-pro-preview reviewer answers the atomic binary checks generated for each prompt with visual evidence retrieval; a sample counts only if it passes every check of the track; averaged over intrinsic-reasoning (abstract prompt) and instruction-following (dense scientific reasoning chain prompt) samples; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 11 models tracked.

Top models

#ModelScore
1Nano Banana Pro90
2Seedream 4.547
3GPT-Image-141
4Qwen-Image-251215
5FLUX-Kontext-Max9
6Show-o2-7B8
7HiDream-I1-Full2
8FLUX.1-dev1
9Stable Diffusion 3.5 Large0
10BAGEL-7B-MoT0

Interactive version: theaggregate.ai/benchmark?slug=sciir-bench-text · How It Works · Data refreshed daily, snapshot 2026-09-29.