AIEvals

Benchmarks

HealthBench Professional (length-adjusted)

Professional healthcare response quality after Anthropic’s response-length adjustment, with deployed safety classifiers enabled.

HealthcareReported by AnthropicVendor-reported onlyReported by Anthropic

How it is run

Sections 8.1 and 8.11 report max-effort mean scores over five runs using Claude Opus 4.8 as judge; the adjustment follows the Sonnet 5.5 system card.

What it cannot tell you

Raw and length-adjusted scoring are distinct. Haiku’s lower effort runs and max runs took place on different days, which limits timing comparisons.

Why there is no independent table

No independent suite in this dataset runs HealthBench Professional (length-adjusted). The numbers below are the ones model publishers report about their own models, which is a different kind of evidence: each publisher chooses its own harness and effort level, so a row is readable on its own and not against the other rows.

Read this with

Other versions of the same evaluation. Versions are not comparable: the task set changes between them.

  • HealthBench Professional (raw)Same family. Raw professional healthcare rubric scores before Anthropic’s response-length correction is applied.
  • HealthBench (length-adjusted)Same category. Quality of healthcare responses under a rubric score adjusted to account for response length rather than raw scoring alone.
  • HealthBench (raw)Same category. Raw healthcare response quality before the length adjustment used in Anthropic’s HealthBench reporting.
  • PhysicianBenchSame category. Successfully carrying out 100 physician tasks inside an electronic health record, measured by the share of attempts passed.

As reported by each publisher

Kept separate per publisher, because the settings are not shared.

Anthropic

Claude Haiku 5.5: system card capability evaluations · 7 October 2026

RankModelScore
3Claude Haiku 5.564.8%
4Claude Haiku 4.532.2%
1Claude Sonnet 5.569.2%
2GPT-6 Luna65.5%

Ranks compare reported scores in the same published row; ties share a rank.

HealthBench Professional (length-adjusted). §8.11.3 / Table 8.1.A / Figure 8.11.1.A: five runs; low/medium/high/xhigh 57.9/59.9/61.3/61.0%. Opus 4.8 judge, classifiers on. Thirteen repeat max runs average 64.2%.

Section 8, pages 111–139 of the Haiku 5.5 system card, read 8 October 2026. Values come from exact prose, tables and labelled figures. Unless specified, Haiku uses adaptive thinking at max effort and five trials with evaluation-specific context up to 1M. Each row retains its subset, metric, effort and source notes; no score is assigned to an unspecified SWE-Bench Pro version. Terminal runs use safeguards without fallback and no internet egress. ProgramBench uses filtered tasks and hidden-test pass rate without a time limit. OSWorld partial and strict results use the official 82-task offline subset. CAD uses 0–1 voxel IoU. Healthcare raw and length-adjusted scores stay separate. Life-science results disable Haiku biology safeguards. Independently run figures quoted by Anthropic remain publisher reports here. Missing comparison values are not inferred from rounded charts.

www.anthropic.com/claude-haiku-5-5-system-card

Sources

Every page used for this benchmark, read on 17 October 2026.