FIKA-Bench - Real-Life: leaderboard

The real-life split of FIKA-Bench (2026), photos from volunteers everyday scenes, answered by closed-book models from the image and question alone. Strict accuracy in percent from an LLM judge that accepts only answers at the required granularity; higher is better.

Metric: Real-life split average, strict answer accuracy (%) on FIKA-Bench, open-ended fine-grained recognition questions (311 images from decontaminated public fine-grained datasets and real-life photos, each answer backed by human-verified evidence URLs, kept after strong closed-book models failed); a text-only LLM judge counts only answers at the required granularity as correct, closed-book models (no tools); higher is better. Source: arxiv.org. Saturation forecast: Around July 2027. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Kimi K2.631#192
2Gemini 3.1 Flash Lite26.7#371
3Qwen 3.5 397B A17B25.9#264
4GPT-5 Mini20.7#364
5Qwen 3 VL 235B A22B20.7#485
6GLM-5V Turbo20.7#360
7Qwen 3.5 9B15.5#828
8Qwen 3 VL 8B12.9#917
9Fine-R1-7B12.1
10VisualRFT-7B5.2

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

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