tensor.news
Integrity IndexModelsCompareLabsPapersMethodologyAbout
tensor.news

© 2026

Evidence

  • Integrity Index
  • Model Claim Ledger
  • Compare

Entities

  • Models
  • Benchmarks
  • Labs
  • Papers

Method

  • Methodology
  • About
  • Privacy

Machine

  • Feed (Atom)
  • llms.txt
  • llms-full.txt
  • Sitemap

tensor.news grades benchmarks and tags model claims by evidence status. Every fact carries a source record and provenance. Scores are framed as task performance under a disclosed harness, never as deployed capability.

← Models

Claude Opus 4.8

Anthropic·released 2026-05-28

independently reproduced

Scored on 18 benchmarks, leading 1. 33% of its scores are independently reproduced — the rest are self-reported or unverified.

18
Benchmarks scored
1
SOTA leads
98.33
Best score
33%
Reproduced
6 of 18

Leads on: OSWorld 2.0

Benchmark fingerprint

Every score, ranked, colored by evidence status.

  • OTIS Mock AIME 2024-2025
    98.33
  • ARC-AGI
    92.5
  • GPQA diamond
    88.05
  • WeirdML
    82.89
  • FrontierMath-Tiers-1-3-v2-Private
    80
  • ARC-AGI-2
    72.08
  • GBAEval
    70.9
  • CursorBench
    63.8
  • SimpleBench
    57.76
  • FrontierMath-Tier-4-v2-Private
    56.1
  • DeepResearch Bench
    50.23
  • APEX-Agents
    42.5
  • SimpleQA Verified
    39.5
  • PostTrainBench
    34.08
  • Chess Puzzles
    34
  • CritPt
    20.86
  • OSWorld 2.0
    20.6
  • Remote Labor Index
    8.33

All benchmark scores

BenchmarkScoreEvidence statusMeasured
OTIS Mock AIME 2024-202598.33accuracy (%)reproduced· optimized2026-05-28
ARC-AGI92.5accuracy (%)unverified· optimized2026-05-28
GPQA diamond88.05accuracy (%)reproduced· optimized2026-05-28
WeirdML82.89accuracy (%)unverified2026-05-28
FrontierMath-Tiers-1-3-v2-Private80accuracy (%)reproduced2026-05-28
ARC-AGI-272.08accuracy (%)unverified2026-05-28
GBAEval70.9accuracy (%)unverified2026-05-28
CursorBench63.8accuracy (%)unverified2026-05-28
SimpleBench57.76accuracy (%)unverified2026-05-28
FrontierMath-Tier-4-v2-Private56.1accuracy (%)reproduced2026-05-28
DeepResearch Bench50.23accuracy (%)unverified2026-05-28
APEX-Agents42.5accuracy (%)unverified2026-05-28
SimpleQA Verified39.5accuracy (%)reproduced2026-05-28
PostTrainBench34.08accuracy (%)unverified2026-05-28
Chess Puzzles34accuracy (%)reproduced2026-05-28
CritPt20.86accuracy (%)unverified2026-05-28
OSWorld 2.020.6accuracy (%)unverified2026-05-28
Remote Labor Index8.33accuracy (%)unverified2026-05-28
18 source records · epoch_benchmarks. Every score is evidence-linked.

Follow the record

More from Anthropic

  • Claude Fable 599.72
  • Claude Opus 4.797.8
  • Claude Sonnet 4.597.73
  • Claude Haiku 4.596.36
  • Claude Opus 4.694.44
  • Claude 3.7 Sonnet91.16

Benchmarks it leads

  • OSWorld 2.0