Claude 3.5 Sonnet (v2) by Anthropic costs $3.00 per 1 Million input tokens and $15.00 per 1 Million output tokens. It features a 200k context window and has a blended 3:1 production rate of $6.000/1M tokens.
Claude 3.5 Sonnet (v2)
Industry benchmark for software engineering, computer use and nuanced long-form comprehension.
Verified Specifications & Benchmark Data
| Developer / Provider | Anthropic |
| Model Family | Claude 3.5 |
| Blended 3:1 Rate (Production Benchmark) | $6.000 / 1M tokens |
| Batch API Discount (24hr SLA) | 50% off standard rate |
| Multimodal Vision | ✅ Supported (Images, Diagrams, OCR) |
| Function Calling / Structured Outputs | ✅ Native Tool Calling |
| MMLU Benchmark Score | 88.3% |
| Latency & Throughput Tier | Fast (~50 t/s) |
| Optimal Architecture & Use Cases | Autonomous coding, complex tool use, multi-step agent reasoning |
Frequently Asked Questions about Claude 3.5 Sonnet (v2)
How much does Claude 3.5 Sonnet (v2) cost per 1M tokens?
Claude 3.5 Sonnet (v2) pricing is set at $3.00 per 1 million input tokens and $15.00 per 1 million output tokens. For high-volume batch workloads, 24-hour batch queues provide a 50% off standard rate.
How much money does Claude 3.5 Sonnet (v2) prompt caching save?
With prompt caching enabled, cached input tokens are discounted to $0.300/1M, saving 90% on repeated system prompts and document vectors.
What is the context window limit of Claude 3.5 Sonnet (v2)?
Claude 3.5 Sonnet (v2) has a maximum context window of 200k (200,000 tokens), supporting up to 8,192 completion tokens per response.
Direct Matchups with Claude 3.5 Sonnet (v2)
See how Claude 3.5 Sonnet (v2) compares against other leading frontier and open-weights models in cost and latency.
Compare Anthropic vs xAI pricing, context limits, and cost per 1M tokens.
Compare Anthropic vs OpenAI pricing, context limits, and cost per 1M tokens.
Compare Anthropic vs Cohere pricing, context limits, and cost per 1M tokens.
Compare Anthropic vs Together AI (Meta) pricing, context limits, and cost per 1M tokens.