Anthropic

claude-opus-4-6

Made by Anthropic. Released 2026-02-05. Accessibility not recorded. We hold no price for this model: no seller we can read lists it any more, so what follows is what was measured, not what it cost.

What it scores, by house

BenchmarkHouseScoreEvaluatedScaffold
chess puzzlesEpoch AI0.1±0.032026-08-06max
chess puzzlesEpoch AI0.1±0.032026-02-20120K
chess puzzlesEpoch AI0.1±0.032026-02-0664K
chess puzzlesEpoch AI0.2±0.042026-02-0632K
ebr benchEpoch AI0.12026-06-29max
frontiermathEpoch AI0.4±0.032026-02-12max
frontiermathEpoch AI0.4±0.032026-02-0632K
frontiermathEpoch AI0.4±0.032026-02-0664K
frontiermathEpoch AI0.4±0.032026-02-05
frontiermath tier 4Epoch AI0.2±0.062026-02-12max
frontiermath tier 4Epoch AI0.2±0.062026-02-0664K
frontiermath tier 4Epoch AI0.2±0.062026-02-0632K
frontiermath tier 4Epoch AI0.1±0.052026-02-05
frontiermath tier 4 v2Epoch AI0.3±0.072026-06-11max
frontiermath tiers 1 3 v2Epoch AI0.7±0.032026-06-11max
gpqa diamondEpoch AI0.9±0.022026-08-06max
gpqa diamondEpoch AI0.9±0.022026-02-0664K
gpqa diamondEpoch AI0.9±0.022026-02-0632K
mystery game puzzlesEpoch AI0.1±0.042026-08-06
mystery game puzzlesEpoch AI0.1±0.032026-08-06low
mystery game puzzlesEpoch AI0.3±0.042026-07-25max
otis mock aime 2024 2025near its ceilingEpoch AI0.9±0.042026-08-06max
otis mock aime 2024 2025near its ceilingEpoch AI0.9±0.032026-02-0632K
otis mock aime 2024 2025near its ceilingEpoch AI0.9±0.032026-02-0664K
simpleqa verifiedEpoch AI0.5±0.022026-08-27max
swe bench verifiedEpoch AI0.8±0.022026-02-18
swe bench verifiedEpoch AI0.8±0.022026-02-10
ale benchcollected by Epoch AI997
algotunecollected by Epoch AI1.5high
apex agentscollected by Epoch AI0.5unknown
apex agentscollected by Epoch AI0.5max
arc aginear its ceilingcollected by Epoch AI0.9120K
arc agi 2collected by Epoch AI0.7120K
cl benchcollected by Epoch AI0.2unknown
cl bench lifecollected by Epoch AI0.2high
cl bench lifecollected by Epoch AI0.1unknown
cybenchcollected by Epoch AI0.9unknown
deepresearchbenchcollected by Epoch AI0.5low
deepresearchbenchcollected by Epoch AI0.5medium
deepresearchbenchcollected by Epoch AI0.6high
dtbenchnear its ceilingcollected by Epoch AI0.9max
enigma evalcollected by Epoch AI0.1
enigma evalcollected by Epoch AI0.1max
forecastbenchcollected by Epoch AI60.0unknown
frontiercodecollected by Epoch AI0.3claude-code
gbaevalcollected by Epoch AI0.4unknown
gsocollected by Epoch AI0.4OpenHands
hlecollected by Epoch AI0.3max
hlecollected by Epoch AI0.2
lmcacollected by Epoch AI55.8max
metr time horizonscollected by Epoch AI719unknown
proofbenchcollected by Epoch AI0.5max
rlicollected by Epoch AI0.0unknown
simplebenchcollected by Epoch AI0.7
terminalbenchcollected by Epoch AI0.7Droid
terminalbenchcollected by Epoch AI0.7TongAgents
terminalbenchcollected by Epoch AI0.7MAYA-V2
terminalbenchcollected by Epoch AI0.7Terminus-KIRA
terminalbenchcollected by Epoch AI0.6Claude Code
terminalbenchcollected by Epoch AI0.8Capy
terminalbenchcollected by Epoch AI0.8Meta-Harness
terminalbenchcollected by Epoch AI0.8ForgeCode
terminalbenchcollected by Epoch AI0.6Terminus 2
terminalbenchcollected by Epoch AI0.7Crux
terminalbenchcollected by Epoch AI0.7Mux
vending bench 2collected by Epoch AI8018unknown
webdev arenacollected by Epoch AI1556unknown
webdev arenacollected by Epoch AI1546high
webdev arenacollected by Epoch AI1536
weirdmlcollected by Epoch AI0.8unknown
weirdmlcollected by Epoch AI0.7
weirdmlcollected by Epoch AI0.8high

These scores are not comparable between houses and are not added up. A score is dated by when it was evaluated, not when the model was released, and a score without its scaffold is not a measurement — which is why both are printed.

What retires, and when

  • deprecation2027-02-05 (Active)

Every rate we hold · Why there is no single best model ·