Compare models

Context, limits, capabilities, price and use cases side by side — up to 4 models at a time. The URL holds your selection, so you can share it.

Claude Haiku 4.5Claude Sonnet 5 Add a modelChat with these 2

AttributeHide identical rows	Claude Haiku 4.5anthropic/claude-haiku-4-5	Claude Sonnet 5anthropic/claude-sonnet-5
Specs
Context window	200K	1M
Max output	64K	128K
Input modalities	TextImage	TextImage
Output modalities	Text	Text
Capabilities
Streaming	Supported	Supported
Tool calling	Supported	Supported
Structured output (JSON)	Supported	Supported
Image input	Supported	Supported
Thinking mode	Supported	Supported
Price / 1M tokens
Input	$ 1.00	$ 2.00
Output	$ 5.00	$ 10.00
Cache read	$ 0.10	$ 0.20
Cache write	$ 1.25	$ 2.50
What it is good at
Best for	Real-time, low-latency tasks High-volume workloads	Agentic coding High-volume agent workloads
Published benchmarks
Terminal-Benchpass_rate	41%(setup not fully disclosed)	—
Terminal-Bench 2.1pass_rate	—	80.4%(setup not fully disclosed)
SWE-Bench Proresolved_rate	—	63.2%(setup not fully disclosed)
SWE-Bench Verifiedresolved_rate	73.3%(setup not fully disclosed)	—
OSWorldsuccess_rate	50.7%(setup not fully disclosed)	—
OSWorld-Verifiedsuccess_rate	—	81.2%(setup not fully disclosed)
GDPval-AA v2elo	—	1618(setup not fully disclosed)
AIME 2025accuracy	80.7%(setup not fully disclosed)	—
MMMLUaccuracy	83%(setup not fully disclosed)	—
MMMUaccuracy	73.2%(setup not fully disclosed)	—
Humanity's Last Examaccuracy	—	43.2%(setup not fully disclosed)
Humanity's Last Exam (with tools)accuracy	—	57.4%(setup not fully disclosed)
GPQA Diamondaccuracy	73%(setup not fully disclosed)	—
τ2-bench Retailscore	83.2%(setup not fully disclosed)	—
τ2-bench Telecomscore	83%(setup not fully disclosed)	—