Sources
This page lists the document behind every score on the leaderboard. Scores are compiled from published documents: OpenAI system cards and update reports, other vendors' model cards, and named leaderboards. Arcophos does not currently run this benchmark itself. Entries are grouped by document. Under each document title sit the rows read from it, each with the score as that document prints it, the sentence or table cell it came from, where in the document it sits, who reported it, and how far the reading has been verified. A row that a second document repeats or disputes appears under that document too, so one model can show up more than once. Rows with no located document are listed at the end. The reading rules are on the methodology page.
Addendum to GPT-6 Astra System Card: GPT-6.1 Sol
- GPT-6.1 Sol36.2
“EVALUATION | GPT-5.5 | GPT-5.6 SOL | GPT-5.6 TERRA | GPT-5.6 LUNA | GPT-6 ASTRA | GPT-6 SOL | GPT-6 LUNA | GPT-6.1 SOL HealthBench Hard length-adjusted | 31.5 (33.8, 2289) | 33.1 (31.1, 1751) | 32.7 (34.3, 2199) | 32.0 (31.4, 1923) | 36.6 (34.2, 1697) | 30.1 (22.1, 974) | 31.4 (25.4, 1241) | 36.2 (33.4, 1646)”
Addendum to GPT-6 Astra System Card: GPT-6.1 Terra
- GPT-6.1 Terra31.4
“EVALUATION GPT-5.5 GPT-5.6-SOL GPT-5.6-TERRA GPT-5.6-LUNA GPT-6 ASTRA GPT-6-SOL GPT-6-LUNA GPT-6.1-SOL GPT-6.1-TERRA HealthBench Hard length-adjusted 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923) 36.6 (34.2, 1697) 30.1 (22.1, 974) 31.4 (25.4, 1241) 36.2 (33.4, 1646) 31.4 (26.0, 1306)”
Baichuan-M3 Technical Report
- GPT-5.2-High42.0
“Notably, in the challenging HealthBench Hard subset, Baichuan-M3 further extends its lead with a score of 44.4, significantly outperforming strong competitors such as GPT-5.2-High (42.0) and AntAngelMed (39.6).”
GPT-5 System Card
- GPT-546.2
“State of the art on HealthBench Hard improves from 31.6% for OpenAI o3 to 46.2% for gpt-5-thinking.”
GPT-5.3 Instant System Card
- GPT-5.3 Chat25.9%
“Hard 26.8% 25.9%”
GPT-5.5 Instant System Card
- GPT-5.3 Chat20.2
“HealthBench Hard 21.6 (23.0, 2,181) 23.3 (23.5, 2,022) 20.2 (17.8, 1,693) 22.9 (21.3, 1,794)”
- GPT-5.5 Instant22.9
“HealthBench Hard 21.6 (23.0, 2,181) 23.3 (23.5, 2,022) 20.2 (17.8, 1,693) 22.9 (21.3, 1,794)”
GPT-5.5 System Card
- GPT-534.7
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289)”
GPT-5.6 - August Updates (system card addendum)
“HealthBench Hard 20.2 (17.8, 1,693) 22.9 (21.3, 1,794) 31.4 (27.1, 1,450) 28.7 (24.9, 1,523)”
“HealthBench Hard 20.2 (17.8, 1,693) 22.9 (21.3, 1,794) 31.4 (27.1, 1,450) 28.7 (24.9, 1,523)”
- GPT-5.3 Chat20.2
“HealthBench Hard 20.2 (17.8, 1,693) 22.9 (21.3, 1,794) 31.4 (27.1, 1,450) 28.7 (24.9, 1,523)”
- GPT-5.5 Instant22.9
“HealthBench Hard 20.2 (17.8, 1,693) 22.9 (21.3, 1,794) 31.4 (27.1, 1,450) 28.7 (24.9, 1,523)”
GPT-5.6 Preview System Card
- GPT-534.7
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.234.3
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.6 Sol33.1
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.6 Terra32.7
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.6 Luna32.0
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.531.5
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.429.1
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.125.4
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
GPT-5.6 System Card
- GPT-534.7
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.234.3
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.6 Sol33.1
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.6 Terra32.7
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.6 Luna32.0
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.531.5
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.429.1
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
- GPT-5.125.4
“HealthBench Hard length-adjusted 34.7 (41.6, 2880) 25.4 (41.4, 4049) 34.3 (38.9, 2585) 29.1 (30.3, 2161) 31.5 (33.8, 2289) 33.1 (31.1, 1751) 32.7 (34.3, 2199) 32.0 (31.4, 1923)”
GPT-6 Astra System Card
- GPT-6 Astra36.3
“Astra has a length-adjusted HealthBench Professional score of 63.4 (+2.9 relative to GPT-5.6 Sol), HealthBench score of 58.1 (+1.1), HealthBench Hard score of 36.3 (+3.2), and HealthBench Consensus score of 95.8 (+0.3).”
GPT-6 Astra System Card - HealthBench (Deployment Safety Hub)
- GPT-6 Astra36.6
“Evaluation | GPT-5.5 | GPT-5.6 Sol | GPT-5.6 Terra | GPT-5.6 Luna | GPT-6 Astra | GPT-6 Sol | GPT-6 Luna HealthBench Hard length-adjusted | 31.5 (33.8, 2289) | 33.1 (31.1, 1751) | 32.7 (34.3, 2199) | 32.0 (31.4, 1923) | 36.6 (34.2, 1697) | 30.1 (22.1, 974) | 31.4 (25.4, 1241)”
- GPT-6 Astra36.3
“HealthBench Hard score of 36.3 (+3.2)”
- GPT-6 Luna31.4
“Evaluation | GPT-5.5 | GPT-5.6 Sol | GPT-5.6 Terra | GPT-5.6 Luna | GPT-6 Astra | GPT-6 Sol | GPT-6 Luna HealthBench Hard length-adjusted | 31.5 (33.8, 2289) | 33.1 (31.1, 1751) | 32.7 (34.3, 2199) | 32.0 (31.4, 1923) | 36.6 (34.2, 1697) | 30.1 (22.1, 974) | 31.4 (25.4, 1241)”
- GPT-6 Sol30.1
“Evaluation | GPT-5.5 | GPT-5.6 Sol | GPT-5.6 Terra | GPT-5.6 Luna | GPT-6 Astra | GPT-6 Sol | GPT-6 Luna HealthBench Hard length-adjusted | 31.5 (33.8, 2289) | 33.1 (31.1, 1751) | 32.7 (34.3, 2199) | 32.0 (31.4, 1923) | 36.6 (34.2, 1697) | 30.1 (22.1, 974) | 31.4 (25.4, 1241)”
gpt-oss-120b & gpt-oss-20b Model Card
- GPT OSS 120B30.0
“HealthBench Hard 22.8 26.9 30.0 9.0 12.9 10.8”
- GPT OSS 20B10.8
“HealthBench Hard 22.8 26.9 30.0 9.0 12.9 10.8”
Introducing Muse Spark: Scaling Towards Personal Superintelligence
- Muse Spark42.8
“HealthBench Hard | Open-Ended Health Queries | 42.8 | 14.8 | 20.6 | 40.1 | 20.3 (image table row; columns Muse Spark Thinking, Opus 4.6 Max, Gemini 3.1 Pro High, GPT-5.4 Xhigh, Grok 4.2 Reasoning)”
Muse Spark Eval Methodology
- Muse Spark42.8
“HealthBench Hard | Open-Ended Health Queries | 42.8 | 14.8 | 20.6 | 40.1 | 20.3 (image table row; columns Muse Spark Thinking, Opus 4.6 Max, Gemini 3.1 Pro High, GPT-5.4 Xhigh, Grok 4.2 Reasoning)”
Rows without a documented source
Every row on the board currently has a located document.
The same fields ship in the JSON download: a source object on every model plus the sources list this page is rendered from. Corrections go to Arcophos.