HELM MedHELM - MIMIC-IV Billing Code: leaderboard

HELM MedHELM task assigning ICD billing codes from MIMIC-IV clinical notes (F1).

Metric: MIMICBillingF1. Source: crfm.stanford.edu. Saturation forecast: Around August 2028. 13 models tracked.

Top models

#ModelScoreOverall rank
1GPT-542.9#98
2GPT-5 Mini (2025-08-07)36.11#172
3Claude 3.7 Sonnet (20250219)35.55#201
4Claude 3.5 Sonnet (20241022)35.29#291
5GPT-4o (2024-05-13)34.59#325
6DeepSeek R129.9#251
7O4 Mini (2025-04-16)29.54#179
8GPT-4o Mini (2024-07-18)26.32#558
9Gemini 2.5 Pro (Preview 05-06)26.21#147
10O3 Mini (2025-01-31)24.29#240
11Gemini 2.0 Flash23.23#342
12Llama 3.3 70B Instruct19.71#534
13Gemini 1.5 Pro (001)18.61#444

No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.

Interactive version: theaggregate.ai/benchmark?slug=helm-medhelm-mimic-iv-billing-code · How It Works · Data refreshed daily, snapshot 2026-10-20.