ModelsCompareBest forBenchmarksStatusPricingAPI

Claude Sonnet 4.5

Anthropic·Operational·Released 2025-09-29textvision
Context200k
Max output64k
Input / 1M$3 ↗
Output / 1M$15 ↗
Cached in$0.3
Benchmarks21

Benchmarks

Cited public sources · rank is exact-benchmark, apples-to-apples

AIME 2025 (MathArena)#15/2384.2% ↗
FrontierMath Tier 4 v2 (Epoch AI run)#39/442.4% ↗
FrontierMath Tiers 1-3 v2 (Epoch AI run)#48/5923.9% ↗
GPQA Diamond (Epoch AI run)#50/7782.3% ↗
OTIS Mock AIME 2024-2025 (Epoch AI run)#55/7677.8% ↗
SimpleQA Verified (Epoch AI run)#40/5330.7% ↗
SWE-bench Verified (Epoch AI run)#15/2471.3% ↗
ARC-AGI-2 (Verified)#7/713.6% ↗
GDPval-AA#2/21276 ↗
GPQA Diamond#39/7083.4% ↗
Humanity's Last Exam (no tools)#22/2917.7% ↗
Humanity's Last Exam (with tools)#11/1333.6% ↗
MCP Atlas#8/943.8% ↗
MMMLU#2/489.5% ↗
MMMU-Pro (no tools)#6/663.4% ↗
MMMU-Pro (with tools)#7/768.9% ↗
OSWorld-Verified#11/1261.4% ↗
SWE-bench Verified#11/4577.2% ↗
Tau2-bench Retail#3/386.2% ↗
Tau2-bench Telecom#3/798% ↗
Terminal-Bench 2.0#8/1351% ↗

Compare

vs GPT-6 Astravs GPT-5.6 Solvs GPT-5.6 Terravs GPT-5.6 Lunavs Claude Fable 5.1vs Claude Opus 5vs Claude Sonnet 5vs Claude Haiku 4.5

More from Anthropic

Claude Fable 5.1Claude Opus 5Claude Sonnet 5Claude Haiku 4.5Claude Opus 4.8Claude Opus 4.7

Access Claude Sonnet 4.5 and every other model through one endpoint with automatic failover: Respan gateway.