Cache read
90% discount on input tokens. A Sonnet 4.6 cache read costs $0.30/M instead of $3.00/M through RunAPI.
LLM API Pricing
Claude API Pricing in 2026
Anthropic charges $3–$15 per million tokens depending on model. RunAPI gives you every Claude model at half the official rate — same API, same output, 50% off your bill.
Updated June 18, 2026
RunAPI Editorial
At a glance
What does the Claude API cost right now?
Anthropic bills in three model tiers: Haiku for lightweight tasks, Sonnet for everyday workloads, and Opus for maximum capability. All prices below are per million tokens. A million tokens is roughly 750,000 words — context usage adds up faster than most people expect.
Cheapest option
Haiku 4.5 at $1/M input and $5/M output through RunAPI. Official rate is $2/$10.
Most popular
Sonnet 4.6 at $3/M input and $15/M output through RunAPI. Official rate is $6/$30.
Maximum capability
Opus 4.7 at $5/M input and $25/M output through RunAPI. Official rate is $10/$50.
Cache discount
Cache reads cost 90% less than standard input tokens — $0.10/M for Haiku through RunAPI. If you're sending the same system prompt on every request, caching can cut your bill dramatically.
Model-by-model breakdown
How much does each Claude model cost per million tokens?
The table below shows official Anthropic pricing alongside RunAPI pricing. RunAPI applies a flat 50% discount across all Claude models. No volume commits, no subscriptions.
| Model | Official input /M | Official output /M | RunAPI input /M | RunAPI output /M | Context window |
|---|---|---|---|---|---|
| Opus 4.8 | $5.00 | $25.00 | $2.50 | $12.50 | 200K |
| Opus 4.7 | $5.00 | $25.00 | $2.50 | $12.50 | 200K |
| Opus 4.6 | $5.00 | $25.00 | $2.50 | $12.50 | 200K |
| Sonnet 4.6 | $6.00 | $30.00 | $3.00 | $15.00 | 200K |
| Sonnet 4.5 | $6.00 | $30.00 | $3.00 | $15.00 | 200K |
| Haiku 4.5 | $2.00 | $10.00 | $1.00 | $5.00 | 200K |
Prompt caching
How do cache discounts reduce your Claude API bill?
Anthropic's prompt caching stores repeated prefixes and charges less when the cached version is reused. This matters for coding agents like Claude Code, which send the same system prompt and file context on every request.
Cache write (5 min TTL)
25% surcharge on input — $3.75/M for Sonnet 4.6 through RunAPI. The cached prefix stays available for 5 minutes.
Cache write (1 hour TTL)
2x the input rate — $6.00/M for Sonnet 4.6 through RunAPI. Useful for long coding sessions where context reuse is frequent.
Batch processing
Anthropic offers a 50% discount on all models for batch requests that tolerate up to 24-hour turnaround. RunAPI passes this discount through.
Real-world costs
What does the Claude API actually cost for real workloads?
Token costs look abstract until you attach them to real tasks. Below are five common developer workloads with estimated monthly costs at two usage levels. Note that input tokens — your system prompt, conversation history, files you attach — often cost more than output tokens for context-heavy workflows.
| Workload | Model | Light use (~50 tasks/day) | Heavy use (~200 tasks/day) | Monthly saving vs official |
|---|---|---|---|---|
| Vibe coding session (Claude Code) | Sonnet 4.6 | $45/mo | $180/mo | $45–$180 |
| PR code review agent | Opus 4.7 | $75/mo | $300/mo | $75–$300 |
| RAG-powered docs chatbot | Haiku 4.5 | $12/mo | $48/mo | $12–$48 |
| Content generation pipeline | Sonnet 4.6 | $30/mo | $120/mo | $30–$120 |
| Multi-agent orchestrator | Opus 4.7 | $150/mo | $600/mo | $150–$600 |
Provider comparison
Is Claude API cheaper than OpenAI and Gemini?
Developers who hit their Claude limits often look at GPT-5 and Gemini 2.5 Pro as alternatives. Here is how the flagship models compare on actual per-million-token costs.
| Provider | Flagship model | Input /M | Output /M | RunAPI rate |
|---|---|---|---|---|
| Anthropic | Claude Opus 4.7 | $10.00 | $50.00 | $5.00 / $25.00 |
| OpenAI | GPT-5.4 | $2.50 | $15.00 | $1.25 / $7.50 |
| Gemini 2.5 Pro | $1.25 | $10.00 | $0.63 / $5.00 |
RunAPI applies a 50% discount on all providers listed above. Prices verified June 2026.
Subscription vs API
Is Claude API cheaper than a Claude Max subscription?
Claude Max costs $100/month for heavy Claude Code usage (or $200 for the 5x plan). The API charges per token with no monthly commitment. One thing the subscription doesn't tell you: the Pro plan has both a daily limit AND a weekly cap, so if you hit it hard every day, you can run out mid-week and wait 3 days for a reset. The API has no such ceiling. For developers running under 10 million output tokens per month on Sonnet 4.6, the RunAPI route costs less than the Max subscription.
Claude Max ($100/mo)
Unlimited usage of Sonnet and limited Opus in Claude Code. Fixed monthly cost. No API access for third-party tools.
Claude API via RunAPI
Pay per token with no monthly commitment. Sonnet 4.6 at $3/M input and $15/M output. $100 buys roughly 6.7 million output tokens — enough for most individual developers.
When Max wins
Heavy daily users who consistently exceed 10 million output tokens per month. The break-even point on Sonnet 4.6 through RunAPI is around 6.7M output tokens.
When API wins
Teams, CI pipelines, multi-model setups, and any tool that requires API access (OpenClaw, n8n, Cursor). No cap, no waitlist, no subscription lock-in. You can also route simpler tasks to Haiku 4.5 — it's about 20x cheaper per token than Sonnet for tasks that don't need full capability.
Getting started
How to access Claude API through RunAPI
1
Create a RunAPI account
Sign up at runapi.ai. No credit card required for the free tier.
2
Copy your API key
Go to Dashboard → API Keys. Create a key and save it — you will use this as your OpenAI API key.
3
Point your SDK to RunAPI
Set the base URL to https://runapi.ai/v1 and use your RunAPI API key. Any OpenAI-compatible client works — Python, Node.js, Go, Ruby, or curl.
4
Start making requests
Use claude-sonnet-4-6, claude-opus-4-7, or any Claude model ID in the model parameter. RunAPI handles routing and billing at 50% of the official rate.
Frequently asked questions
Claude API Pricing FAQ
I asked 1 question and used 20% of my 5-hour usage. What's going on?
This happens because large context windows — your system prompt, conversation history, attached files — dominate token cost. One question with a big context can burn 20% of your 5-hour window. The API lets you control exactly what context you send on each request, so you stop paying for context you don't need.
Does RunAPI modify Claude's output?
Using third-party tools like OpenClaw, n8n, or Cursor with a personal Claude subscription can get your account banned. Anthropic's terms don't allow subscription plans to power external apps. The API is the only supported path — and RunAPI makes it pay-as-you-go with no monthly commitment.
Can I use RunAPI with Claude Code?
Yes. Set ANTHROPIC_BASE_URL to https://runapi.ai and your RunAPI key as the API key. Claude Code works without any other modification. You pay per token at 50% of the official rate — no subscription required.
What happens if Anthropic changes their pricing?
RunAPI adjusts within 24 hours. The 50% discount is maintained relative to Anthropic's published rates.
Is there a free tier?
Yes. New accounts get free credits to test any model. After that, billing is strictly pay-as-you-go with no minimum.
Does RunAPI support prompt caching?
Yes. Cache reads, 5-minute writes, and 1-hour writes are all supported at 50% of Anthropic's cache pricing. If you send the same large system prompt on every request, cache reads cut that input cost by 90% — from $3.00/M to $0.30/M on Sonnet 4.6.
Can I use the OpenAI SDK to call Claude through RunAPI?
Yes. RunAPI is OpenAI-compatible. Point any OpenAI client at runapi.ai/v1 and use Claude model IDs. If you're already calling Claude from OpenRouter or a similar gateway, switching to RunAPI works the same way — you change one URL and one API key.
How does billing work?
Pay-as-you-go. You fund your account with a balance, and each API call deducts the token cost. No subscriptions, no invoices, no contracts. You can also set a hard spend limit in your RunAPI dashboard so you never get a surprise bill — unlike direct Anthropic billing, which has caught users off guard.
RELATED
Explore More
OpenAI API Pricing
GPT-5.4, GPT-5.4-mini, and o3 pricing with RunAPI vs official rates.
Gemini API Pricing
Gemini 2.5 Pro and Flash pricing — the cheapest flagship model.
Claude Code Pricing
What Claude Code costs with API billing vs Max subscription.
All Model Pricing
Compare 220+ models across text, image, video, and audio.
Start using Claude at half price.
Create a free RunAPI account, get your API key, and start calling Claude Opus, Sonnet, or Haiku at 50% off official Anthropic pricing.