Anthropic
•
Listed September 2025
•
Multimodal Foundation
Claude Sonnet 4.5
Arena #64 • score 1456 ±3
Proprietary model from Anthropic, listed September 2025. Arena text score 1456 (#64 of 414 entries). $3.00 input / $15.00 output per 1M tokens. 1M tokens context.
Parameters
Not disclosed
Context Window
1M tokens
Input / 1M tokens
$3.00
Output / 1M tokens
$15.00
Benchmark Results
- Arena Text Elo: 1456 (#64 of 414 on arena.ai · ±3 · 83,973 votes · variant: claude-sonnet-4-5-20250929-high-32k)
- OSWorld-Verified: 62.9% (Success rate · 100 steps · tested 2025-10-31)
- SWE-bench Verified: 71.4% (mini-SWE-agent (bash only) · run 'Claude 4.5 Sonnet (high)' · 2026-02-17)
- AA Intelligence Index: 20.7 (Artificial Analysis v4.3 · best setting: Claude 4.5 Sonnet (Reasoning) · checked 2026-10-11)
- AA Coding Index: 52.1 (Artificial Analysis v4.3 · best setting: Claude 4.5 Sonnet (Reasoning) · checked 2026-10-11)
- AA Agentic Index: 15.8 (Artificial Analysis v4.3 · best setting: Claude 4.5 Sonnet (Reasoning) · checked 2026-10-11)