ReceiptBench: leaderboard

Visual information extraction on ReceiptBench (2026), structured extraction of 19 fields from photos and scans of real receipts and invoices (purchase, dining, travel, hotel, taxi and more), 2,000 held-out test images, JSON output. Overall F1 over all fields. Field-level F1 times 100 (0-100), exact, numeric or list matching with a small LLM judge for near misses; higher is better.

Metric: Overall field-level F1 (times 100, so 0-100) on the 2,000-image held-out test set of ReceiptBench (real travel and expense receipts and invoices, 19 annotated fields), one-shot schema prompt with JSON output; fields are scored by exact, numeric or structured-list matching with a Qwen3-4B semantic judge when string similarity falls below the match threshold, over all 19 fields; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 14 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Pro73.73#145
2Qwen 3 VL Plus72.1#512
3GPT-570.76#178
4Qwen 3 VL 32B67.51#622
5InternVL3.5-241B67.42#629
6Qwen 3 VL 8B65.45#917
7InternVL3-78B64.43#759
8PaddleOCR-VL63.44
9Qwen 3 VL 4B62.61#1151
10olmOCR-7B62.28
11InternVL3-8B57.72#1782
12DianJin-OCR-R149.79
13DeepSeek-OCR-small39.59
14InternVL3-2B38.07#3290

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

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