Anthropic · Pricing & cost calculator

Claude 3.5 Sonnet pricing

Mid-2024 flagship still used heavily. Sonnet 4 is newer but similarly priced.

Input tokens

$0.00300

per 1,000 tokens

Output tokens

$0.01500

per 1,000 tokens

Context window

200,000

tokens max

Prices sourced from Anthropic's public API rate card. Last reviewed April 2026.

Calculate your Claude 3.5 Sonnet spend

Drag the sliders. The math runs in your browser.

Per request

$0.012000

at 1,500/500 tokens

Monthly spend

$120

10,000 requests × $0.012000

Could save ~73% with a similar-quality alternative

Switch from Claude 3.5 Sonnet to Claude 3.5 Haiku and save $88.00/month

Same workload at 1,500/500 tokens × 10,000 req/month: $120 $32. ~4× cheaper on input.

TrackCost's gateway auto-detects which requests are safe to route to the cheaper model and flags the ones that need Claude 3.5 Sonnet's full capability. You stay in control — we surface the opportunity.

When should you use Claude 3.5 Sonnet?

  • Production chat
  • Long-context RAG

Claude 3.5 Sonnet pricing FAQ

How much does Claude 3.5 Sonnet cost per 1,000 tokens?

Claude 3.5 Sonnet costs $0.00300 per 1,000 input tokens and $0.01500 per 1,000 output tokens as of the current Anthropic API pricing.

What is a cheaper alternative to Claude 3.5 Sonnet?

Claude 3.5 Haiku is significantly cheaper than Claude 3.5 Sonnet (~4× cheaper on input) and often works fine for tasks that don't need Claude 3.5 Sonnet's full capability. TrackCost automatically suggests model downgrades where safe.

How can I reduce my Claude 3.5 Sonnet API costs?

The three highest-leverage ways: (1) route requests to a cheaper model where possible, (2) cache identical requests so you don't pay twice for the same response, (3) set hard monthly budgets that block requests when exceeded. TrackCost automates all three at the gateway layer.

What's the context window for Claude 3.5 Sonnet?

Claude 3.5 Sonnet supports up to 200,000 tokens of context. For comparison with other models, use the TrackCost savings calculator.

Compare pricing across other models