AIEvals

Benchmarks

WANDR (frozen index)

Collecting many entities, verifying their facts and citations, and balancing precision and recall with soft F1 across 500 tasks.

Wide researchReported by AnthropicVendor-reported onlyReported by Anthropic

How it is run

Section 8.8.3 uses offline search and fetch against a frozen index, a 980k-token budget, and Claude Opus 4.6 judging; malformed entity rows are dropped.

What it cannot tell you

Perplexity uses live web tools and a different judge. These modified-harness results must remain separate from the original WANDR leaderboard.

Why there is no independent table

No independent suite in this dataset runs WANDR (frozen index). 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

Nothing else on this site shares this benchmark's family or category, so these are other evaluations the same publisher reports.

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 5.549.9%
4Claude Haiku 4.56.4%
2Claude Sonnet 5.570.0%
1Claude Opus 5.572.3%

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

WANDR (frozen index). Figure 8.8.3.A labelled max soft-F1 scores. Haiku low/medium/high/xhigh: 3.5/12.9/37.3/45.9%. Frozen search index, Opus 4.6 judge, 980k task budget.

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.