Self-hosting: unknown
On-premise: unknown
Edge deployability: unknown
API availability: Available on Claude Platform; also available in Claude Code
Regions available: unknown

Benchmarks

as reported — vendor-supplied unless otherwise noted

Artificial Analysis Intelligence Index (max effort) 43CursorBench Not specifiedFrontierCode 1.1 58.4%

National-security pathway & evidence

from the assessed source item, when one is linked

Impact: mediumConfidence: lowNat-sec: noneanalysis

Evidence limitations: This is a single analysis-source account; benchmark and product claims are attributed to Anthropic, Artificial Analysis, and other vendors, and the source notes some cost figures are provisional.

[AINews] Claude Haiku 5.5 — better than GPT-6 Luna at the same pricing

Linked research papers

from paper_model_links

StoreBench: A Live-Commerce Environment for Evaluating and Training Autonomous Operator Agentsevaluates · 80% confidence
When Interfaces Speak: Data-Aware Generative UI Harness for Active Interactionevaluates · 95% confidence
Mine Odyssey: Benchmarking Spatial Agentic Intelligence in the Wildevaluates · 100% confidence