Maternity Case Record Digitisation - Formatted Fields: leaderboard

A 2026 benchmark of multimodal LLMs for digitising a handwritten South African maternity case record: each form image is extracted into a JSON template and compared field by field with reference JSON files. This board is exact-match accuracy in percent on populated formatted fields such as dates and numerical values, median over the real registry forms; higher is better.

Metric: Accuracy on populated formatted fields (dates in DD/MM/YYYY and numerical values with units; from 0-1, times 100; median over forms), exact match after normalisation, on the real redacted Maternity Case Records (49 forms for Gemini and OpenAI models; 46 for Claude) of the UBOMI BUHLE Pregnancy Exposure Registry (South Africa), one form image per call with the optimised extraction prompt and JSON template, temperature 0, two attempts at valid JSON; higher is better. Source: arxiv.org. Saturation forecast: Not forecast. 9 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Claude Sonnet 4.677#173
2Claude Haiku 4.573#583
3GPT-5 Mini72#364
4Gemini 3.1 Flash Lite72#371
5Gemini 3.1 Pro (Preview)67#99
6GPT-5.439#135
7Claude Opus 4.636#113
8Gemini 2.5 Flash32#463
9Gemini 2.5 Pro27#276

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

Interactive version: theaggregate.ai/benchmark?slug=maternity-case-record-digitisation-formatted-fields · How It Works · Data refreshed daily, snapshot 2026-11-21.