Atlas

Models

← All models

Sonnet 4.5

Anthropic · 2025-09-29

Claude Sonnet 4.5 is Anthropic's hybrid-reasoning model for coding, agentic tasks, and computer use. It supports optional extended thinking; at launch, Anthropic reported it as state of the art on SWE-bench Verified and observed it maintaining focus for more than 30 hours on complex, multi-step tasks.

Benchmark scores

BenchmarkScore
A-Fantasia40.3
A-Fantasia Backwards Spelling24.0
A-Fantasia Chess25.0
A-Fantasia Cube Rotation72.0
AA-LCR65.7
AA-Omniscience Index0.3
Agent Red Teaming k=100 (CAIS)64.6
AIME 2024 (avg@8)90.7
AIME 2024-202588.2
AIME 202537.0
AIME 2025 (avg@8)85.7
ALE-Bench1192
AlgoTune1.5
Apex (MathArena)1.6
APEX-SWE31.0
ARC-AGI-163.7
ARC-AGI-213.6
Arena Code WebDev — Brand & Marketing1384
Arena Code WebDev — Consumer Product1384
Arena Code WebDev — Content Creation Tools1384
Arena Code WebDev — Data & Analytics1387
Arena Code WebDev — Gaming1391
Arena Code WebDev — HTML1390
Arena Code WebDev — React1384
Arena Code WebDev — Reference-Based Design1388
Arena Code WebDev — Simulations1383
Arena Document (Style Controlled)1446
Arena Text — Business, Management & Financial Ops (No Style Control)1439
Arena Text — Chinese (No Style Control)1464
Arena Text — Creative Writing (No Style Control)1441
Loading Atlas data…