Skip to main content

Anthropic AI Models

AI safety company creating helpful, harmless, and honest AI systems

11Models
$0.80Starting from
Try Anthropic
$0.80
Cheapest Input Price
Claude 3.5 Haiku
$6.254545
Average Input Price
per 1M tokens
$15.00
Most Expensive
Claude Opus 4

Anthropic Models & Pricing

ModelInput PriceOutput PriceContext WindowMax OutputCategoryCapabilities
Claude Sonnet 4.6
Improved Sonnet with frontier-level coding, computer use, and reasoning at balanced pricing
Released: 2/18/2026
$3.00
per 1M tokens
$15.00
per 1M tokens
1,000,000
tokens
65,536
tokens
balanced
textvisioncodereasoningcomputer-use
Claude Opus 4.6
Most intelligent model for building agents and coding
Released: 2/5/2026
$5.00
per 1M tokens
$25.00
per 1M tokens
1,000,000
tokens
128,000
tokens
flagship
textvisionreasoningcode
Claude Opus 4.5
Latest Opus iteration with improved performance, released November 2025
Released: 11/1/2025
$5.00
per 1M tokens
$25.00
per 1M tokens
200,000
tokens
64,000
tokens
flagship
textvisioncodereasoning
Claude Haiku 4.5
Fastest model with near-frontier intelligence
Released: 10/15/2025
$1.00
per 1M tokens
$5.00
per 1M tokens
200,000
tokens
64,000
tokens
efficient
textvision
Claude Sonnet 4.5
Best combination of speed and intelligence
Released: 9/29/2025
$3.00
per 1M tokens
$15.00
per 1M tokens
200,000
tokens
64,000
tokens
balanced
textvisioncode
Claude Opus 4.1
Enhanced Claude 4 Opus with improved performance, released August 2025
Released: 8/5/2025
$15.00
per 1M tokens
$75.00
per 1M tokens
200,000
tokens
32,000
tokens
flagship
textvisioncodereasoning
Claude Opus 4
Anthropic's highly capable model released May 2025 with advanced reasoning
Released: 5/14/2025
$15.00
per 1M tokens
$75.00
per 1M tokens
200,000
tokens
32,000
tokens
flagship
textvisioncodereasoning
Claude Sonnet 4
Balanced performance and speed, released May 2025
Released: 5/14/2025
$3.00
per 1M tokens
$15.00
per 1M tokens
200,000
tokens
64,000
tokens
standard
textvisioncodereasoning
Claude 3.5 Haiku
Fast and affordable model for everyday tasks
Released: 11/4/2024
$0.80
per 1M tokens
$4.00
per 1M tokens
200,000
tokens
8,192
tokens
efficient
textvision
Claude 3.5 Sonnet
Previous generation balanced model with strong coding abilities
Released: 10/22/2024
$3.00
per 1M tokens
$15.00
per 1M tokens
200,000
tokens
8,192
tokens
balanced
textvisioncode
Claude 3 Opus
Previous generation flagship with exceptional reasoning
Released: 3/4/2024
$15.00
per 1M tokens
$75.00
per 1M tokens
200,000
tokens
4,096
tokens
flagship
textvisionreasoning

Calculate Anthropic Costs

Use our calculator to estimate costs for any Anthropic model based on your usage patterns.

Compare Anthropic Models

Anthropic Pricing FAQ

How much does Anthropic API cost?

Anthropic offers 11 models with input pricing ranging from $0.80 to $15.00 per million tokens. The most affordable option is Claude 3.5 Haiku at $0.80/1M input and $4.00/1M output tokens.

What is the cheapest Anthropic model?

Claude 3.5 Haiku is the cheapest Anthropic model at $0.80 per million input tokens and $4.00 per million output tokens. It supports a 200,000-token context window.

Which Anthropic model should I use?

It depends on your use case. Claude 3.5 Haiku is best for cost-sensitive applications. Claude Opus 4 ($15.00/1M input) offers the strongest capabilities. For high-volume workloads, their efficient-tier models offer the best cost-performance ratio.

What models does Anthropic offer?

Anthropic currently offers 11 models: Claude Opus 4.6, Claude Sonnet 4.6, Claude Sonnet 4.5, Claude Opus 4, Claude Sonnet 4, Claude Opus 4.1, Claude Opus 4.5, Claude Haiku 4.5, Claude 3.5 Sonnet, Claude 3.5 Haiku, Claude 3 Opus. These span flagship, balanced, standard, efficient categories with context windows up to 1000K tokens.

How does Anthropic pricing compare to competitors?

Anthropic's pricing starts at $0.80/1M input tokens with Claude 3.5 Haiku. Use our comparison tool to see side-by-side pricing against other providers like OpenAI, Google, and Mistral.

About Anthropic

AI safety company creating helpful, harmless, and honest AI systems