ModelPricing.aiAnthropic Claude API Pricing
Anthropic Claude API pricing spans from lightweight Haiku models to the flagship Fable 5 / Mythos 5 lineup. Below is the current standard pricing breakdown for Claude models tracked by ModelPricing.ai, including prompt-cache read and write rates. Rates sourced from the official Anthropic pricing page. Compare with OpenAI and Google Gemini pricing, or see all models in our LLM pricing comparison.
Anthropic Claude API Pricing at a Glance
These are the Claude rows most teams compare first: Haiku for low-cost routing, Sonnet for balanced work, Opus for deep reasoning, and Fable 5 / Mythos 5 for the most demanding long-horizon work.
| Model | Input / 1M | Output / 1M | Cache Read / 1M | Best Fit |
|---|---|---|---|---|
| Claude Haiku 4.5 | $1.00 | $5.00 | $0.10 | Fast routing and support flows |
| Claude Sonnet 4.6 | $3.00 | $15.00 | $0.30 | Coding, analysis, and agents |
| Claude Opus 4.8 | $5.00 | $25.00 | $0.50 | Deep reasoning and research |
| Claude Fable 5 / Mythos 5 | $10.00 | $50.00 | $1.00 | Long-horizon agentic work |
All Claude Models
| Model | Input $/1M tokens | Output $/1M tokens | Type | Notes |
|---|---|---|---|---|
| claude-3-7-sonnet | $3.00 | $15.00 | Flat | |
| claude-fable-5 | $10.00 | $50.00 | Flat | |
| claude-haiku-3 | $0.250 | $1.25 | Flat | |
| claude-haiku-3-5 | $0.800 | $4.00 | Flat | |
| claude-haiku-4-5 | $1.00 | $5.00 | Flat | |
| claude-mythos-5 | $10.00 | $50.00 | Flat | |
| claude-opus-3 | $15.00 | $75.00 | Flat | |
| claude-opus-4-0 | $15.00 | $75.00 | Flat | |
| claude-opus-4-1 | $15.00 | $75.00 | Flat | |
| claude-opus-4-5 | $5.00 | $25.00 | Flat | |
| claude-opus-4-6 | $5.00 | $25.00 | Flat | |
| claude-opus-4-7 | $5.00 | $25.00 | Flat | |
| claude-opus-4-8 | $5.00 | $25.00 | Flat | |
| claude-sonnet-4-0 | $3.00 | $15.00 | Flat | |
| claude-sonnet-4-5 | $3.00 | $15.00 | Flat | |
| claude-sonnet-4-6 | $3.00 | $15.00 | Flat |
How Claude Pricing Works
Claude models use flat pricing — a fixed rate per million tokens for both input and output, rather than breakpoint pricing. Fable 5 / Mythos 5, Opus 4.8, Opus 4.7, Opus 4.6, and Sonnet 4.6 include the full 1M context window at standard rates.
Anthropic also offers prompt caching: frequently reused context (system prompts, long documents) can be cached so subsequent reads cost just 10% of the input rate. Cache writes cost 1.25× input at 5-minute TTL or 2× input at 1-hour TTL. For applications that reuse the same context across many requests, caching can reduce input costs by 90% after the first write.
Anthropic also documents batch discounts, data-residency uplifts, and fast mode pricing for selected recent models. The estimator table here focuses on standard token and cache rates.
Claude Model Comparison
Haiku
Fastest and most affordable. Ideal for classification, extraction, and high-volume tasks where speed matters more than reasoning depth.
Sonnet
Best balance of intelligence and cost. Great for coding assistance, analysis, and multi-step workflows that need strong reasoning.
Opus, Fable, and Mythos
Premium Claude tiers for complex reasoning, research, and high-autonomy work. Fable 5 / Mythos 5 sit above Opus 4.8 at the $10/$50 standard rate.
Claude API Monthly Cost Estimates
How much will Claude API cost you per month? Here are realistic estimates based on typical usage patterns. Use our LLM cost calculator for a precise estimate based on your specific workload.
Light Use
$5-20/mo
- Personal projects
- <1K requests/day
- Haiku for most tasks
Medium Use
$20-100/mo
- Small team apps
- 1-5K requests/day
- Mix of Haiku and Sonnet
Heavy Use
$100-500/mo
- Production apps
- 5-20K requests/day
- Sonnet for quality tasks
Enterprise
$500+/mo
- Large-scale deployments
- 20K+ requests/day
- Opus or Fable 5 / Mythos 5 for complex tasks
Which Claude Model Should You Use?
Choosing the right Claude model depends on your task complexity and budget. Here's a quick guide based on common use cases.
| Use Case | Recommended Model | Est. Monthly Cost | Why This Model |
|---|---|---|---|
| Customer support chatbot | Haiku 4.5 | $10-50 | Fast responses, affordable for high volume |
| Code generation | Sonnet 4.6 | $30-150 | Best balance of code quality and cost |
| Research & analysis | Opus 4.8 | $100-500 | Deepest reasoning, handles complex tasks |
| Data extraction | Haiku 3 | $5-20 | Cheapest option, sufficient for structured tasks |
| Agentic workflows | Sonnet 4.6 | $50-200 | Strong reasoning at mid-tier pricing |
4 Ways to Reduce Your Claude API Costs
Use model routing
Start with Haiku for triage and only escalate to Sonnet or Opus when the task requires deeper reasoning. This can cut costs 50-80% for mixed workloads.
Use prompt caching aggressively
Cache reads cost just 10% of the input rate. If your system prompt, tool definitions, or context documents repeat across requests, mark them as cacheable — the break-even is a single reuse at 5-minute TTL.
Optimize output length
Output tokens cost 5x more than input tokens across all Claude models. Set appropriate max_tokens limits and use structured output formats to keep responses concise.
Monitor usage with the ModelPricing API
Track your token consumption programmatically to identify cost spikes and optimize your model selection over time. Get started free.
How Claude Pricing Compares to GPT and Gemini
Claude sits in the mid-to-premium tier compared to OpenAI and Google Gemini. Haiku is competitive for lightweight work, Sonnet is the balanced Claude tier, and Opus is the premium reasoning tier. See our full LLM pricing comparison for all models.
| Tier | Claude | OpenAI | |
|---|---|---|---|
| Budget | Haiku 3 — $0.25/$1.25 | GPT-5 Nano — $0.05/$0.40 | Flash Lite — $0.075/$0.30 |
| Mid-range | Sonnet 4.6 — $3/$15 | GPT-5 — $1.25/$10 | Gemini 2.5 Pro — $1.25/$10 |
| Flagship | Fable 5 / Mythos 5 — $10/$50 | GPT-5.4 — $2.50/$15 | Gemini 3.1 Pro — $2/$12 |
Frequently Asked Questions
How much does Claude API cost?
Claude API pricing varies by model. Claude Haiku 3 starts at $0.25/1M input tokens, Claude Opus 4.8 costs $5/1M input and $25/1M output tokens, and Claude Fable 5 / Mythos 5 cost $10/1M input and $50/1M output tokens. Most teams spend $20-200/month depending on usage volume and model selection.
What is the cheapest Claude model?
Claude Haiku 3 is the most affordable at $0.25/1M input tokens and $1.25/1M output tokens, making it ideal for high-volume, low-complexity tasks like classification and data extraction.
Does Claude have breakpoint pricing?
Claude standard API pricing is flat per model, not breakpoint-based. Fable 5 / Mythos 5, Opus 4.8, Opus 4.7, Opus 4.6, and Sonnet 4.6 include the 1M token context window at standard rates. US-only inference, fast mode, batch, and prompt caching can still change the effective price.
How much does Claude Code cost?
Claude Code uses the Claude API under the hood. Costs depend on which model you configure — Sonnet 4.6 at $3/$15 per million tokens is the current balanced option. Heavy Claude Code users typically spend $50-200/month on API usage. Anthropic also offers Max subscription plans as an alternative to per-token billing.
Is the Claude API free?
Anthropic API access is usage-based. Any trial credits, promos, or account-specific grants can change, so check your Claude Console billing page before relying on free usage.
Claude API vs subscription — which is cheaper?
For light usage, the Claude Pro subscription ($20/month) is often cheaper than API billing. For heavy or programmatic usage, the API gives you more control. Claude Max plans ($100-200/month) can be significantly cheaper than equivalent API usage for power users.
How does Claude prompt caching affect pricing?
Anthropic prompt caching charges 1.25x base input for 5-minute cache writes, 2x for 1-hour writes, and 0.1x for cache hits. That makes reused system prompts, tools, or documents much cheaper after the initial write.
Claude Sonnet vs Opus — which should I use?
Sonnet 4.6 ($3/$15 per million tokens) is the best choice for most tasks including coding, analysis, and multi-step workflows. Opus 4.8 ($5/$25) is worth the premium for complex reasoning and research. Fable 5 / Mythos 5 ($10/$50) are the highest-cost Claude models for the most demanding long-horizon work.
Estimate Your Claude API Costs
Get accurate, real-time cost estimates for any Claude model with our API. Or try the LLM cost calculator to compare across all providers.
Get Started Free