Claude Opus 4.8 vs Claude Opus 5

$4.08 / task
Wins 0 of 17 benchmarks
$2.19 / task
Wins 12 of 17 benchmarks
Different efforts: max · medium · Compare all at high
AutomationBench-AA AutomationBench-AA: Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation
Opus 4.845.6
Opus 554.3
GDPval-AA (work tasks) 2.1 GDPval-AA (work tasks) 2.1: Real work tasks from many professions (reports, spreadsheets, analyses), graded against expert work
Opus 4.847.7
Opus 549.6
τ³-bench · banking · 1.0.1 τ-bench official · 3-banking-1.0.1: Customer-service agents converse with simulated users, retrieve domain knowledge and change state through tools
Opus 4.839.7
Opus 5 · max48.7
τ³-Banking · AA τ³-Banking · AA: Agents retrieve banking knowledge and execute tools to resolve customer-support workflows
Opus 4.834.2
Opus 538.6
AA Analyst Agent AA Analyst Agent: Data-analyst tasks: the model explores data with tools and reports findings
Opus 4.845
Opus 5 · max53.8
Terminal-Bench 4.0 Terminal-Bench 4.0 · source setting 2: Agents use a terminal to complete executable coding, system administration and research tasks in the 4.0 suite
Opus 4.821.7
Opus 534.3
AA-Briefcase 1.1 · Elo AA-Briefcase 1.1 · Elo: Agents produce professional deliverables; blind pairwise comparisons produce the source Elo rating
Opus 4.81,321.4
Opus 51,434.9
AA-LCR (long context) 1.1 AA-LCR (long context) 1.1: Questions that need reading and connecting facts across very long documents (around 100K tokens)
Opus 4.877.7
Opus 582
DeepSWE 1.1 DeepSWE 1.1: Agents change repository code to resolve software engineering tasks under the stated DeepSWE edition
Opus 4.859
Opus 568.9
EnterpriseOps-Gym-AA EnterpriseOps-Gym-AA: Agents complete stateful enterprise workflows in eight domains; final database state determines success
Opus 4.844
Opus 5 · max47.5
EBR Bench EBR Bench: Agents learn to play Earthborne Rangers across repeated playthroughs
Opus 4.828.6
Opus 5 · max45.7
Furniture Assembly Furniture Assembly: Models inspect furniture assembly photos, identify incorrect builds and locate the mistaken step
Opus 4.842.5
Opus 5 · max60.8
Toolathlon-Verified Toolathlon-Verified: Toolathlon-Verified as reported in this launch evaluation; the post is the authority for the recorded task and grading scope
Opus 4.876.2
Opus 5—
Terminal-Bench Hard (version unstated) Terminal-Bench Hard (version unstated): Agents solve difficult software, system-administration and data-processing tasks in a terminal
Opus 4.858.3
Opus 5—
Terminal-Bench Science 0.1 Terminal-Bench Science 0.1: Agents complete expert-authored research workflows across five scientific domains in a terminal
Opus 4.8—
Opus 5 · max28.6
SWE-bench Pro (v2 full) SWE-bench Pro (v2 full): Agents resolve repository issues under the network-locked v2 protocol; full and 51-task Hard results are separate
Opus 4.8—
Opus 5 · xhigh99.4
SWE-bench Pro (v2 hard) SWE-bench Pro (v2 hard): Agents resolve repository issues under the network-locked v2 protocol; full and 51-task Hard results are separate
Opus 4.8—
Opus 5 · xhigh98

Specs

SpecOpus 4.8Opus 5
Input, per 1M$5$5
Output, per 1M$25$25
Cost per task?$4.08$2.19
Speed, tok/s6655
Context1M1M
WeightsClosedClosed
ReleasedMay 2026Jul 2026
See images✓✓
Read PDFsPartialPartial
Hear audio——
Watch video——
Call tools✓✓
Force a tool callPartialPartial
Return strict JSON✓✓
Search the web??
Run code??
Use a computer??

What breaks when you switch

Opus 4.8 → Opus 5
Nothing known.
Opus 5 → Opus 4.8
Nothing known.