HealthBench Professional

OpenAI logoGPT-5 on HealthBench Professional

Rank 36 of 43, updated November 2, 2026.

GPT-5 scores 0.462 on HealthBench Professional, rank 36 of 43 models on the board. OpenAI's August 2025 flagship, still served at unchanged prices and now three generations back. 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.

36
rank of 43
0.462
length-adjusted score
0.510
score before the length adjustment
0.241
gap to the leader, GPT-6 Astra (Anthropic run)
400K
context window, tokens
$1.25 / $10.00
input / output price per 1M tokens

Where it sits on the board

The models ranked around GPT-5, by length-adjusted score. Differences are against GPT-5.

RankModelLength-adjustedBefore the adjustmentDifference
33DDeepSeek V4.1 Flash0.5070.529+0.045
34SpaceX AI logoGrok 4.60.485–+0.023
35OpenAI logoGPT-5.40.4810.519+0.019
36OpenAI logoGPT-50.4620.510–
37OpenAI logoGPT-5.20.4590.500-0.003
38Anthropic logoClaude Sonnet 4.60.442–-0.020
39OpenAI logoGPT-5.6 Luna (August)0.4410.468-0.021

Result and API facts

Rank36 of 43
Length-adjusted score0.462
Score before the length adjustment0.510
LabOpenAI
Context window400K tokens
API price per 1M tokens$1.25 in / $10.00 out
Licenseproprietary
SourceGPT-5.6 System Card (system card)
Released2025-08-07

Source of this score

Read from GPT-5.6 System Card (system card, OpenAI, 2026-07-09). Vendor-reported. Confidence: verified. Configuration: length-adjusted, max reasoning effort, GPT-5.6 system card Table 6 column GPT-5 (51.0 unadjusted, 3616 chars).

Section 5.1 HealthBench, Table 6 (reported as length-adjusted score (unadjusted, mean response length in characters)), column GPT-51 corroborating documentfull entry on the sources page
HealthBench Professional length-adjusted 46.2 (51.0, 3616) 39.6 (48.0, 4863) 45.9 (50.0, 3400) 48.1 (51.9, 3308) 51.8 (57.2, 3818) 60.5 (64.1, 3228) 57.7 (62.4, 3618) 55.7 (59.8, 3389)

Position in the field

The gap to the leader, GPT-6 Astra (Anthropic run) at 0.703, is 0.241. Directly above sits GPT-5.4 at 0.481. Directly below sits GPT-5.2 at 0.459. 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-5 score on HealthBench Professional?

GPT-5 scores 0.462 on HealthBench Professional, which places it at rank 36 of 43 models on the board as of November 2, 2026. The number was read from GPT-5.6 System Card, listed on the sources page.

How much does GPT-5 cost to run?

GPT-5 is priced at $1.25 per million input tokens and $10.00 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-5 vs GPT-6 Astra (Anthropic run)
    0.462 vs 0.703, GPT-6 Astra (Anthropic run) by 0.241
  • GPT-5 vs Claude Sonnet 5.5
    0.462 vs 0.692, Claude Sonnet 5.5 by 0.230
  • GPT-5 vs Claude Fable 5
    0.462 vs 0.660, Claude Fable 5 by 0.198
  • GPT-5 vs Claude Opus 5.5
    0.462 vs 0.656, Claude Opus 5.5 by 0.194
  • GPT-5 vs GPT-6 Luna (Anthropic run)
    0.462 vs 0.655, GPT-6 Luna (Anthropic run) by 0.193
  • GPT-5 vs Claude Haiku 5.5
    0.462 vs 0.648, Claude Haiku 5.5 by 0.186
  • GPT-5 vs GPT-6 Astra
    0.462 vs 0.647, GPT-6 Astra by 0.185
  • GPT-5 vs GPT-6.1 Sol
    0.462 vs 0.642, GPT-6.1 Sol by 0.180
  • GPT-5 vs Claude Fable 5 (September card)
    0.462 vs 0.633, Claude Fable 5 (September card) by 0.171
  • GPT-5 vs GPT-6.1 Terra
    0.462 vs 0.627, GPT-6.1 Terra by 0.165
  • GPT-5 vs Muse Wonder Full
    0.462 vs 0.623, Muse Wonder Full by 0.161
  • GPT-5 vs Claude Fable 5.1
    0.462 vs 0.621, Claude Fable 5.1 by 0.159
  • GPT-5 vs GPT-6 Luna
    0.462 vs 0.608, GPT-6 Luna by 0.146
  • GPT-5 vs GPT-6 Sol
    0.462 vs 0.608, GPT-6 Sol by 0.146
  • GPT-5 vs GPT-5.6 Sol
    0.462 vs 0.605, GPT-5.6 Sol by 0.143
  • GPT-5 vs Muse Wonder
    0.462 vs 0.605, Muse Wonder by 0.143
  • GPT-5 vs Claude Opus 5
    0.462 vs 0.598, Claude Opus 5 by 0.136
  • GPT-5 vs Muse Spark 1.1
    0.462 vs 0.593, Muse Spark 1.1 by 0.131
  • GPT-5 vs Claude Sonnet 5
    0.462 vs 0.578, Claude Sonnet 5 by 0.116
  • GPT-5 vs GPT-5.6 Terra
    0.462 vs 0.577, GPT-5.6 Terra by 0.115
  • GPT-5 vs Claude Opus 4.8 (Opus 4.8 grader)
    0.462 vs 0.574, Claude Opus 4.8 (Opus 4.8 grader) by 0.112
  • GPT-5 vs Grok 4.7
    0.462 vs 0.567, Grok 4.7 by 0.105
  • GPT-5 vs Claude Opus 4.8
    0.462 vs 0.558, Claude Opus 4.8 by 0.096
  • GPT-5 vs GPT-5.6 Luna
    0.462 vs 0.557, GPT-5.6 Luna by 0.095
  • GPT-5 vs Gemini 4 Argon
    0.462 vs 0.551, Gemini 4 Argon by 0.089
  • GPT-5 vs Muse Spark
    0.462 vs 0.541, Muse Spark by 0.079
  • GPT-5 vs GPT-5.6 Sol (August)
    0.462 vs 0.540, GPT-5.6 Sol (August) by 0.078
  • GPT-5 vs Kimi K3
    0.462 vs 0.532, Kimi K3 by 0.070
  • GPT-5 vs GLM 5.3
    0.462 vs 0.528, GLM 5.3 by 0.066
  • GPT-5 vs Claude Opus 4.7
    0.462 vs 0.519, Claude Opus 4.7 by 0.057
  • GPT-5 vs GPT-5.5
    0.462 vs 0.518, GPT-5.5 by 0.056
  • GPT-5 vs Gemini 3.8 Flash
    0.462 vs 0.517, Gemini 3.8 Flash by 0.055
  • GPT-5 vs DeepSeek V4.1 Flash
    0.462 vs 0.507, DeepSeek V4.1 Flash by 0.045
  • GPT-5 vs Grok 4.6
    0.462 vs 0.485, Grok 4.6 by 0.023
  • GPT-5 vs GPT-5.4
    0.462 vs 0.481, GPT-5.4 by 0.019
  • GPT-5 vs GPT-5.2
    0.462 vs 0.459, GPT-5 by 0.003
  • GPT-5 vs Claude Sonnet 4.6
    0.462 vs 0.442, GPT-5 by 0.020
  • GPT-5 vs GPT-5.6 Luna (August)
    0.462 vs 0.441, GPT-5 by 0.021
  • GPT-5 vs GPT-5.1
    0.462 vs 0.396, GPT-5 by 0.066
  • GPT-5 vs GPT-5.5 Instant
    0.462 vs 0.384, GPT-5 by 0.078
  • GPT-5 vs MAI-Thinking-1
    0.462 vs 0.350, GPT-5 by 0.112
  • GPT-5 vs Claude Haiku 4.5
    0.462 vs 0.322, GPT-5 by 0.140

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