Atlas

Benchmarks

← All benchmarks

RCM-Bench-1: Patient billing

Health & Medicine · 2026-10-02

Pass rate over three attempts on each of the 8 RCM-Bench-1 scenarios for explaining balances, applying credits and handling refunds and permissions.

Benchmark source: Commure

Top models (higher is better)

ModelScoreSource
Claude Opus 5.5100.0RCM-Bench-1 results: Claude Opus 5.5, Patient billing · Commure
Claude Opus 5100.0RCM-Bench-1 results: Claude Opus 5, Patient billing · Commure
Claude Sonnet 5.5100.0RCM-Bench-1 results: Claude Sonnet 5.5, Patient billing · Commure
Gemini 3.8 Flash100.0RCM-Bench-1 results: Gemini 3.8 Flash, Patient billing · Commure
Gemini 4 Argon100.0RCM-Bench-1 results: Gemini 4 Argon, Patient billing · Commure
GPT-6 Luna100.0RCM-Bench-1 results: GPT-6 Luna, Patient billing · Commure
GPT-6 Sol100.0RCM-Bench-1 results: GPT-6 Sol, Patient billing · Commure
Claude Haiku 5.595.8RCM-Bench-1 results: Claude Haiku 5.5, Patient billing · Commure
DeepSeek V4.1 Flash95.8RCM-Bench-1 results: DeepSeek V4.1 Flash, Patient billing · Commure
GPT-5.6 Sol95.8RCM-Bench-1 results: GPT-5.6 Sol, Patient billing · Commure
GPT-6.1 Sol95.8RCM-Bench-1 results: GPT-6.1 Sol, Patient billing · Commure
Opus 4.687.5RCM-Bench-1 results: Claude Opus 4.6, Patient billing · Commure
Sonnet 4.687.5RCM-Bench-1 results: Claude Sonnet 4.6, Patient billing · Commure
GLM-5.387.5RCM-Bench-1 results: GLM-5.3, Patient billing · Commure
GPT-5.6 Terra87.5RCM-Bench-1 results: GPT-5.6 Terra, Patient billing · Commure
GPT-5 Mini87.5RCM-Bench-1 results: GPT-5 mini, Patient billing · Commure
Grok 4.787.5RCM-Bench-1 results: Grok 4.7, Patient billing · Commure
Sonnet 579.2RCM-Bench-1 results: Claude Sonnet 5, Patient billing · Commure
GPT-5.6 Luna79.2RCM-Bench-1 results: GPT-5.6 Luna, Patient billing · Commure
Haiku 4.550.0RCM-Bench-1 results: Claude Haiku 4.5, Patient billing · Commure
Loading Atlas data…