Claude API Pricing 2026:
Opus 4.6, Sonnet 4.6 & Haiku 4.5 Costs
Current Anthropic Claude API pricing for Opus 4.6, Sonnet 4.6, and Haiku 4.5. Compare token costs, prompt caching, batch discounts, and long-context pricing. Last verified: 2026-04-01.
Anthropic's Claude pricing in 2026 has three clear tiers: Opus 4.6 for the highest-end reasoning and agentic tasks, Sonnet 4.6 as the general-purpose production default, and Haiku 4.5 for high-volume, cost-sensitive workloads. A key upgrade in 2026: Opus 4.6 and Sonnet 4.6 both support 1 million token context at standard pricing.
Claude API Standard Pricing 2026
| Model | Input / 1M tokens | Output / 1M tokens | Context window | Best for |
|---|---|---|---|---|
| Claude Opus 4.6 | $5.00 | $25.00 | 1M tokens | Highest-end reasoning, agents, complex coding |
| Claude Sonnet 4.6 | $3.00 | $15.00 | 1M tokens | Production default for complex tasks |
| Claude Haiku 4.5 | $1.00 | $5.00 | 200K tokens | High-volume, lower-cost workloads |
Prompt Caching Pricing
Claude's prompt caching delivers a 90% discount on cache reads versus standard input pricing. Cache writes are slightly more expensive (1.25× standard), but for repeated large system prompts or RAG contexts, the savings compound quickly.
| Model | Cache Write / 1M | Cache Read / 1M | Savings vs standard |
|---|---|---|---|
| Claude Opus 4.6 | $6.25 | $0.50 | 90% |
| Claude Sonnet 4.6 | $3.75 | $0.30 | 90% |
| Claude Haiku 4.5 | $1.25 | $0.10 | 90% |
Batch API Pricing
Anthropic's Message Batches API provides a 50% discount for non-real-time workloads (classification, data processing, evals). Batches complete within 24 hours.
| Model | Batch Input / 1M | Batch Output / 1M |
|---|---|---|
| Claude Opus 4.6 | $2.50 | $12.50 |
| Claude Sonnet 4.6 | $1.50 | $7.50 |
| Claude Haiku 4.5 | $0.50 | $2.50 |
Claude vs OpenAI: Quick Price Comparison
| Tier | Claude model | Claude input / 1M | OpenAI model | OpenAI input / 1M |
|---|---|---|---|---|
| Budget | Haiku 4.5 | $1.00 | GPT-4o mini | $0.15 |
| Balanced | Sonnet 4.6 | $3.00 | GPT-5.4 | $2.50 |
| Premium | Opus 4.6 | $5.00 | GPT-5.4 | $2.50 |
| Long context | Opus / Sonnet (1M ctx) | standard | GPT-4o (128K) | — |
GPT-4o mini remains cheaper for simple, high-volume tasks. Claude Opus 4.6 is significantly cheaper than it was in early 2025, making it genuinely competitive for complex reasoning and agent workflows. Claude's 1M-token context on Sonnet and Opus remains a strong advantage for document-heavy applications.
Which Claude Model Should You Choose?
| Use case | Best model | Reason |
|---|---|---|
| High-volume chatbots, classification | Haiku 4.5 | Lowest cost, handles most routine tasks |
| Customer support, summarization | Sonnet 4.6 | Best quality-to-cost ratio, 1M context |
| Complex coding, content creation | Sonnet 4.6 | Production default for demanding tasks |
| Complex agentic workflows, research | Opus 4.6 | Highest accuracy, 1M context, now at $5/M |
| RAG with large corpora | Sonnet 4.6 | 1M context avoids chunking complexity |
Real-World Cost Examples
Customer Support Bot — 50,000 conversations/month
- System prompt: 2,000 tokens (cached after first call)
- Per conversation: 500 tokens fresh input + 300 tokens output
- Sonnet 4.6 without caching: ~$22/month
- Sonnet 4.6 with prompt caching: ~$3–4/month (cache reads at $0.30/M)
- Savings: 85%+ reduction with caching on repeated system prompt
Document Analysis Pipeline — 10,000 docs/month
- Average: 5,000 tokens input + 800 tokens output
- Haiku 4.5: $5 input + $4 output = $9/month
- Sonnet 4.6: $15 input + $12 output = $27/month
Legacy Models
Older Claude models (Claude 3.5 Sonnet, Claude 3 Haiku, Claude 3.7 Sonnet) remain available on the API for existing integrations. For new projects, migrate to the Claude 4 family. Anthropic typically maintains legacy models for 12+ months after newer versions launch.
API Access and Free Tier
- Sign up at console.anthropic.com — no approval process required
- Free tier: limited credits on Claude Haiku 4.5 for testing
- Paid tier: pay-as-you-go, no monthly minimum
- Enterprise: custom pricing for $10,000+/month spend — contact Anthropic sales
Frequently Asked Questions
Is Claude Sonnet 4.6 the best default production model?
For most teams, yes. It delivers strong performance on complex tasks at $3/M input, and the 1M-token context handles large documents without chunking. Haiku 4.5 is the better choice when you need high volume and cost is the primary constraint.
Does Claude batch processing reduce cost?
Yes — Anthropic's Message Batches API gives a 50% discount on all models. Batches accept up to 10,000 requests and complete within 24 hours, making them ideal for classification, evals, and non-latency-sensitive pipelines.
Does Claude charge extra for 1M token context?
No. Opus 4.6 and Sonnet 4.6 support up to 1M tokens in a single context window at standard per-token pricing. There is no surcharge for long-context requests.
When is Haiku 4.5 cheaper than GPT-4o mini in practice?
GPT-4o mini ($0.15/M input) is cheaper per token than Haiku 4.5 ($1.00/M). However, if Haiku produces better output quality for your task — requiring fewer retries or shorter prompts — total cost can be comparable. Benchmark both on your specific use case before choosing.
Which Claude model is best for RAG or coding?
Sonnet 4.6 is the go-to for both. For RAG, its 1M context window can handle large document sets without a retrieval layer. For coding, it handles complex multi-file tasks and follows detailed instructions reliably. Opus 4.6 is worth the premium only for the most complex multi-step reasoning or research tasks.
Calculate Your Claude API Costs
Get an accurate monthly cost estimate for your Claude API usage.
Open API Cost Calculator