Anthropic

Claude Opus 5

Made by Anthropic. Released 2026-07-24. Accessibility not recorded. Cheapest standard rate we hold: $25 per million output tokens, no context split.

What it costs

TierKindInOutCached inFrom
batchtext$2.50$13seller
standardtext$5$25$0.50seller

per million tokens, US dollars. * marks a rate from a single registry or two that disagree — the best available number rather than a confirmed one. Every other row is confirmed against a seller or a licensed index.

What it scores, by house

BenchmarkHouseScoreEvaluatedScaffold
chess puzzlesEpoch AI0.2±0.042026-08-06low
chess puzzlesEpoch AI0.3±0.052026-08-06
chess puzzlesEpoch AI0.4±0.052026-07-24max
ebr benchEpoch AI0.5±0.042026-09-05max
frontiermath tier 4 v2Epoch AI0.7±0.072026-07-24max
frontiermath tiers 1 3 v2Epoch AI0.9±0.022026-07-24max
gpqa diamondEpoch AI0.9±0.022026-08-06low
gpqa diamondEpoch AI0.9±0.022026-08-06
gpqa diamondEpoch AI0.9±0.012026-07-24max
mystery game puzzlesEpoch AI0.4±0.052026-08-06
mystery game puzzlesEpoch AI0.6±0.052026-07-25max
otis mock aime 2024 2025near its ceilingEpoch AI1.0±0.022026-08-06
otis mock aime 2024 2025near its ceilingEpoch AI0.9±0.042026-08-06low
otis mock aime 2024 2025near its ceilingEpoch AI1.0±0.012026-07-24max
simpleqa verifiedEpoch AI0.6±0.022026-08-10max
ale benchcollected by Epoch AI2165high
apex agentscollected by Epoch AI0.6max
arc aginear its ceilingcollected by Epoch AI1.0high
arc aginear its ceilingcollected by Epoch AI1.0max
arc agi 2collected by Epoch AI0.9max
arc agi 2collected by Epoch AI0.9high
blueprint bench 2collected by Epoch AI0.3unknown
btf3collected by Epoch AI0.1FutureSearch agent
critptcollected by Epoch AI0.3xhigh
critptcollected by Epoch AI0.3medium
critptcollected by Epoch AI0.2low
critptcollected by Epoch AI0.3high
critptcollected by Epoch AI0.3max
cursorbenchcollected by Epoch AI0.7xhigh
cursorbenchcollected by Epoch AI0.6medium
cursorbenchcollected by Epoch AI0.7max
cursorbenchcollected by Epoch AI0.7high
cursorbenchcollected by Epoch AI0.6low
deepswecollected by Epoch AI0.7mini-swe-agent
dtbenchnear its ceilingcollected by Epoch AI1.0max
frontiercodecollected by Epoch AI0.5claude-code
gbaevalcollected by Epoch AI0.8unknown
gdp pdfcollected by Epoch AI0.2max
lmcacollected by Epoch AI63.3max
osworld 2collected by Epoch AI0.2low
osworld 2collected by Epoch AI0.3medium
osworld 2collected by Epoch AI0.3high
osworld 2collected by Epoch AI0.3xhigh
osworld 2collected by Epoch AI0.3max
posttrainbenchcollected by Epoch AI0.4Claude Code
proofbenchcollected by Epoch AI1.0max
scicodecollected by Epoch AI0.5medium
scicodecollected by Epoch AI0.6max
scicodecollected by Epoch AI0.5low
scicodecollected by Epoch AI0.5high
scicodecollected by Epoch AI0.5xhigh
simplebenchcollected by Epoch AI0.8unknown
vending bench 2collected by Epoch AI11182unknown
webdev arenacollected by Epoch AI1691max
webdev arenacollected by Epoch AI1663high
weirdmlcollected by Epoch AI0.9max
weirdmlcollected by Epoch AI0.9
weirdmlcollected by Epoch AI0.9high

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

  • deprecation2027-07-24 (Active)

Headlines that name it

Matched by the model's name appearing in our headline, literally. There is no link from a story to a model in our data, so this is a search result rather than a claim that the story is about it.

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