Claude Fable 5 vs Codestral (25.08)

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

Anthropic
Claude Fable 5GA
9.5
AI Panel
Mistral AI
Codestral (25.08)GA
7.6
AI Panel

Identity & lifecycle

ProviderAnthropicMistral AI
Family / tierClaude 5 · FableCodestral · Coder
StatusGAGA
Released2026-06-092025-07-31
Knowledge cutoff2026-012025-05

Architecture & context

Context window1M128K
Max output tokens128K16K
Input modalitiestext, image, filetext
Reasoning modealwaysnone
Open weightsNoNo

Pricing (per Mtok)

Input$10$0.3
Output$50$0.9
Cached input$1$0.03
Batch input$5$0.15
Free tierNoYes

Speed

Output speed (tok/s)——
Time to first token (s)——
Latency tierslowfast

Trust & deployment

Trains on inputsNoNo
LicenseProprietaryProprietary
CloudsBedrock, Vertex AI, Azure AI FoundryBedrock, Azure AI Foundry, Vertex AI
ComplianceSOC2, HIPAA, GDPR, ISO27001SOC2, ISO27001, GDPR

AI Panel scoring

Unified score9.57.6
Decision Maker9.57.5
Domain Strategist9.57.5
Finance Lead78
Domain Practitioner9.58
Power User97.5
Skeptic7.57
Value score6.58.5

Benchmarks

SWE-bench Verified95—
HumanEval—86.6
Terminal-Bench88—
LMArena Elo1506—
LMArena Coding Elo1552—

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