HealthBench Professional

OpenAI logoGPT-6 Luna (Anthropic run) on HealthBench Professional

Rank 5 of 43, updated November 2, 2026.

GPT-6 Luna (Anthropic run) scores 0.655 on HealthBench Professional, rank 5 of 43 models on the board. HealthBench Professional scores models on 525 tasks drawn from real clinician conversations, graded criterion by criterion against physician-written rubrics on a 0 to 1 scale.

5
rank of 43
0.655
length-adjusted score
–
score before the length adjustment, not printed in the source
0.048
gap to the leader, GPT-6 Astra (Anthropic run)
1.05M
context window, tokens
$0.10 / $0.50
input / output price per 1M tokens

Where it sits on the board

The models ranked around GPT-6 Luna (Anthropic run), by length-adjusted score. Differences are against GPT-6 Luna (Anthropic run).

RankModelLength-adjustedBefore the adjustmentDifference
2Anthropic logoClaude Sonnet 5.50.6920.771+0.037
3Anthropic logoClaude Fable 50.6600.703+0.005
4Anthropic logoClaude Opus 5.50.6560.771+0.001
5OpenAI logoGPT-6 Luna (Anthropic run)0.655––
6Anthropic logoClaude Haiku 5.50.6480.710-0.007
7OpenAI logoGPT-6 Astra0.6470.682-0.008
8OpenAI logoGPT-6.1 Sol0.6420.672-0.013

Result and API facts

Rank5 of 43
Length-adjusted score0.655
Score before the length adjustmentnot printed in the source
LabOpenAI
Context window1.05M tokens
API price per 1M tokens$0.10 in / $0.50 out
Licenseproprietary
SourceClaude Haiku 5.5 System Card (system card)
Released2026-09-22

Source of this score

Read from Claude Haiku 5.5 System Card (system card, Anthropic, 2026-10-07). Independent run. Confidence: verified. Configuration: Run by Anthropic through OpenAI's public API; max effort; Opus 4.8 grader; length-adjusted; not OpenAI's published figure.

p. 132, Figure 8.11.1.A, HealthBench Professional panel, GPT-6 Luna (OpenAI) diamond label (visually read printed label); Section 8.11.1 text on pp. 131-132full entry on the sources page
HealthBench Professional panel: 65.5% (GPT-6 Luna (OpenAI)); caption: GPT-6 Luna's scores come from our own runs through OpenAI's public API and are not OpenAI's published figures.

Position in the field

The gap to the leader, GPT-6 Astra (Anthropic run) at 0.703, is 0.048. Directly above sits Claude Opus 5.5 at 0.656. Directly below sits Claude Haiku 5.5 at 0.648. The scores on this page are compiled from published documents rather than from one controlled run, so a small gap between 2 models can reflect a difference in grader version, reasoning effort, or deployment setting as well as a difference in capability.

What does GPT-6 Luna (Anthropic run) score on HealthBench Professional?

GPT-6 Luna (Anthropic run) scores 0.655 on HealthBench Professional, which places it at rank 5 of 43 models on the board as of November 2, 2026. The number was read from Claude Haiku 5.5 System Card, listed on the sources page.

How much does GPT-6 Luna (Anthropic run) cost to run?

GPT-6 Luna (Anthropic run) is priced at $0.10 per million input tokens and $0.50 per million output tokens through OpenAI's API.

Pairings with a dedicated comparison page are linked; every other difference is in the score-difference matrix.

  • GPT-6 Luna (Anthropic run) vs GPT-6 Astra (Anthropic run)
    0.655 vs 0.703, GPT-6 Astra (Anthropic run) by 0.048
  • GPT-6 Luna (Anthropic run) vs Claude Sonnet 5.5
    0.655 vs 0.692, Claude Sonnet 5.5 by 0.037
  • GPT-6 Luna (Anthropic run) vs Claude Fable 5
    0.655 vs 0.660, Claude Fable 5 by 0.005
  • GPT-6 Luna (Anthropic run) vs Claude Opus 5.5
    0.655 vs 0.656, Claude Opus 5.5 by 0.001
  • GPT-6 Luna (Anthropic run) vs Claude Haiku 5.5
    0.655 vs 0.648, GPT-6 Luna (Anthropic run) by 0.007
  • GPT-6 Luna (Anthropic run) vs GPT-6 Astra
    0.655 vs 0.647, GPT-6 Luna (Anthropic run) by 0.008
  • GPT-6 Luna (Anthropic run) vs GPT-6.1 Sol
    0.655 vs 0.642, GPT-6 Luna (Anthropic run) by 0.013
  • GPT-6 Luna (Anthropic run) vs Claude Fable 5 (September card)
    0.655 vs 0.633, GPT-6 Luna (Anthropic run) by 0.022
  • GPT-6 Luna (Anthropic run) vs GPT-6.1 Terra
    0.655 vs 0.627, GPT-6 Luna (Anthropic run) by 0.028
  • GPT-6 Luna (Anthropic run) vs Muse Wonder Full
    0.655 vs 0.623, GPT-6 Luna (Anthropic run) by 0.032
  • GPT-6 Luna (Anthropic run) vs Claude Fable 5.1
    0.655 vs 0.621, GPT-6 Luna (Anthropic run) by 0.034
  • GPT-6 Luna (Anthropic run) vs GPT-6 Luna
    0.655 vs 0.608, GPT-6 Luna (Anthropic run) by 0.047
  • GPT-6 Luna (Anthropic run) vs GPT-6 Sol
    0.655 vs 0.608, GPT-6 Luna (Anthropic run) by 0.047
  • GPT-6 Luna (Anthropic run) vs GPT-5.6 Sol
    0.655 vs 0.605, GPT-6 Luna (Anthropic run) by 0.050
  • GPT-6 Luna (Anthropic run) vs Muse Wonder
    0.655 vs 0.605, GPT-6 Luna (Anthropic run) by 0.050
  • GPT-6 Luna (Anthropic run) vs Claude Opus 5
    0.655 vs 0.598, GPT-6 Luna (Anthropic run) by 0.057
  • GPT-6 Luna (Anthropic run) vs Muse Spark 1.1
    0.655 vs 0.593, GPT-6 Luna (Anthropic run) by 0.062
  • GPT-6 Luna (Anthropic run) vs Claude Sonnet 5
    0.655 vs 0.578, GPT-6 Luna (Anthropic run) by 0.077
  • GPT-6 Luna (Anthropic run) vs GPT-5.6 Terra
    0.655 vs 0.577, GPT-6 Luna (Anthropic run) by 0.078
  • GPT-6 Luna (Anthropic run) vs Claude Opus 4.8 (Opus 4.8 grader)
    0.655 vs 0.574, GPT-6 Luna (Anthropic run) by 0.081
  • GPT-6 Luna (Anthropic run) vs Grok 4.7
    0.655 vs 0.567, GPT-6 Luna (Anthropic run) by 0.088
  • GPT-6 Luna (Anthropic run) vs Claude Opus 4.8
    0.655 vs 0.558, GPT-6 Luna (Anthropic run) by 0.097
  • GPT-6 Luna (Anthropic run) vs GPT-5.6 Luna
    0.655 vs 0.557, GPT-6 Luna (Anthropic run) by 0.098
  • GPT-6 Luna (Anthropic run) vs Gemini 4 Argon
    0.655 vs 0.551, GPT-6 Luna (Anthropic run) by 0.104
  • GPT-6 Luna (Anthropic run) vs Muse Spark
    0.655 vs 0.541, GPT-6 Luna (Anthropic run) by 0.114
  • GPT-6 Luna (Anthropic run) vs GPT-5.6 Sol (August)
    0.655 vs 0.540, GPT-6 Luna (Anthropic run) by 0.115
  • GPT-6 Luna (Anthropic run) vs Kimi K3
    0.655 vs 0.532, GPT-6 Luna (Anthropic run) by 0.123
  • GPT-6 Luna (Anthropic run) vs GLM 5.3
    0.655 vs 0.528, GPT-6 Luna (Anthropic run) by 0.127
  • GPT-6 Luna (Anthropic run) vs Claude Opus 4.7
    0.655 vs 0.519, GPT-6 Luna (Anthropic run) by 0.136
  • GPT-6 Luna (Anthropic run) vs GPT-5.5
    0.655 vs 0.518, GPT-6 Luna (Anthropic run) by 0.137
  • GPT-6 Luna (Anthropic run) vs Gemini 3.8 Flash
    0.655 vs 0.517, GPT-6 Luna (Anthropic run) by 0.138
  • GPT-6 Luna (Anthropic run) vs DeepSeek V4.1 Flash
    0.655 vs 0.507, GPT-6 Luna (Anthropic run) by 0.148
  • GPT-6 Luna (Anthropic run) vs Grok 4.6
    0.655 vs 0.485, GPT-6 Luna (Anthropic run) by 0.170
  • GPT-6 Luna (Anthropic run) vs GPT-5.4
    0.655 vs 0.481, GPT-6 Luna (Anthropic run) by 0.174
  • GPT-6 Luna (Anthropic run) vs GPT-5
    0.655 vs 0.462, GPT-6 Luna (Anthropic run) by 0.193
  • GPT-6 Luna (Anthropic run) vs GPT-5.2
    0.655 vs 0.459, GPT-6 Luna (Anthropic run) by 0.196
  • GPT-6 Luna (Anthropic run) vs Claude Sonnet 4.6
    0.655 vs 0.442, GPT-6 Luna (Anthropic run) by 0.213
  • GPT-6 Luna (Anthropic run) vs GPT-5.6 Luna (August)
    0.655 vs 0.441, GPT-6 Luna (Anthropic run) by 0.214
  • GPT-6 Luna (Anthropic run) vs GPT-5.1
    0.655 vs 0.396, GPT-6 Luna (Anthropic run) by 0.259
  • GPT-6 Luna (Anthropic run) vs GPT-5.5 Instant
    0.655 vs 0.384, GPT-6 Luna (Anthropic run) by 0.271
  • GPT-6 Luna (Anthropic run) vs MAI-Thinking-1
    0.655 vs 0.350, GPT-6 Luna (Anthropic run) by 0.305
  • GPT-6 Luna (Anthropic run) vs Claude Haiku 4.5
    0.655 vs 0.322, GPT-6 Luna (Anthropic run) by 0.333

Where the scores come from and how they are read is on the methodology page. The full ranking is on the leaderboard.