Claude Opus 4.8 vs Claude Fable 5 vs Claude Opus 4.7

Best cell per row highlighted. Null means undisclosed — never counted as zero.

Anthropic
Claude Opus 4.8GA
9.2
AI Panel
Anthropic
Claude Fable 5GA
9.5
AI Panel
Anthropic
Claude Opus 4.7GA
9.0
AI Panel

Identity & lifecycle

ProviderAnthropicAnthropicAnthropic
Family / tierClaude 4 · OpusClaude 5 · FableClaude 4 · Opus
StatusGAGAGA
Released2026-05-272026-06-082026-04-15
Knowledge cutoff2026-01

Architecture & context

Context window1M1M1M
Max output tokens128K128K128K
Input modalitiestext, imagetext, image, filetext, image
Reasoning modealwaysalwaysalways
Open weightsNoNoNo

Pricing (per Mtok)

Input$5$10$5
Output$25$50$25
Cached input$0.5$1$0.5
Batch input$2.5$5$2.5
Free tierNoNoNo

Speed

Output speed (tok/s)6054.6
Time to first token (s)0.523.85
Latency tiermediumslowslow

Trust & deployment

Trains on inputsNoNoNo
LicenseProprietaryProprietaryProprietary
CloudsBedrock, Vertex AI, Azure AI FoundryBedrock, Vertex AI, Azure AI FoundryBedrock, Vertex AI, Azure AI Foundry
ComplianceSOC2, HIPAA, GDPR, ISO27001SOC2, HIPAA, GDPR, ISO27001SOC2, HIPAA, GDPR, ISO27001

AI Panel scoring

Unified score9.29.59
Decision Maker99.59
Domain Strategist99.59
Finance Lead8.578
Domain Practitioner9.59.59.5
Power User998.5
Skeptic87.57.5
Value score86.57.5

Benchmarks

GPQA Diamond93.694.2
SWE-bench Verified88.69587.6
MMMU91.5
Terminal-Bench74.68869.4
Humanity's Last Exam54.7
LMArena Elo1503
LMArena Coding Elo1554
Artificial Analysis Index57

Scores link to their sources. Missing cells mean the vendor hasn't disclosed a result — honesty over padding.