Claude Code for less. Up to 65% below Anthropic's API.
Run Claude Code on a Zurelay key: the same Claude models through Anthropic's own Messages API, with prompt caching and thinking levels, below Anthropic's API list price. Prepaid credit from $10, no subscription.
- Claude models
- 10
- Below API list
- 46–65%
- $100 of Opus becomes
- $35.00
{ "env": { "ANTHROPIC_BASE_URL": "https://api.zurelay.com", "ANTHROPIC_AUTH_TOKEN": "<ZURELAY_API_KEY>", "ANTHROPIC_DEFAULT_OPUS_MODEL": "claude-opus-5-5", "ANTHROPIC_DEFAULT_SONNET_MODEL": "claude-sonnet-5-5", "ANTHROPIC_DEFAULT_HAIKU_MODEL": "claude-haiku-5-5" }, "model": "claude-opus-5-5"}Same session
A smaller bill for the same work. Every token type.
Input, cache writes, cache reads and output are all priced below Anthropic's API list price by the same share, so any mix of them costs that much less. Here's what $100 of Claude Code at list price comes to.
Pricing
Every Claude model. Cache prices included.
Claude Code reads most of its prompt from the cache, so cache prices matter as much as input and output. Each is shown with Anthropic's API list price struck through.
- Input / 1M
- $1.40
$4.00 - Cache write / 1M
- $1.75
$5.00 - Cache read / 1M
- $0.14
$0.40 - Output / 1M
- $7.00
$20.00
- Input / 1M
- $0.90
$2.00 - Cache write / 1M
- $1.125
$2.50 - Cache read / 1M
- $0.09
$0.20 - Output / 1M
- $4.50
$10.00
- Input / 1M
- $0.04
$0.10 - Cache write / 1M
- $0.05
$0.125 - Cache read / 1M
- $0.004
$0.01 - Output / 1M
- $0.20
$0.50
- Input / 1M
- $3.50
$10.00 - Cache write / 1M
- $4.375
$12.50 - Cache read / 1M
- $0.35
$1.00 - Output / 1M
- $17.50
$50.00
- Input / 1M
- $3.50
$10.00 - Cache write / 1M
- $4.375
$12.50 - Cache read / 1M
- $0.35
$1.00 - Output / 1M
- $17.50
$50.00
- Input / 1M
- $1.75
$5.00 - Cache write / 1M
- $2.1875
$6.25 - Cache read / 1M
- $0.175
$0.50 - Output / 1M
- $8.75
$25.00
- Input / 1M
- $2.40
$5.00 - Cache write / 1M
- $3.00
$6.25 - Cache read / 1M
- $0.24
$0.50 - Output / 1M
- $12.00
$25.00
- Input / 1M
- $1.78
$5.00 - Cache write / 1M
- $2.225
$6.25 - Cache read / 1M
- $0.178
$0.50 - Output / 1M
- $8.93
$25.00
- Input / 1M
- $2.40
$5.00 - Cache write / 1M
- $3.00
$6.25 - Cache read / 1M
- $0.24
$0.50 - Output / 1M
- $12.00
$25.00
- Input / 1M
- $0.54
$1.00 - Cache write / 1M
- $0.675
$1.25 - Cache read / 1M
- $0.054
$0.10 - Output / 1M
- $2.70
$5.00
| Model | Input / 1M | Cache write / 1M | Cache read / 1M | Output / 1M | You save |
|---|---|---|---|---|---|
| Claude Opus 5.5claude-opus-5-5 | $1.40 | $1.75 | $0.14 | $7.00 | 65%off |
| Claude Sonnet 5.5claude-sonnet-5-5 | $0.90 | $1.125 | $0.09 | $4.50 | 55%off |
| Claude Haiku 5.5claude-haiku-5-5 | $0.04 | $0.05 | $0.004 | $0.20 | 60%off |
| Claude Fable 5.1claude-fable-5-1 | $3.50 | $4.375 | $0.35 | $17.50 | 65%off |
| Claude Fable 5claude-fable-5 | $3.50 | $4.375 | $0.35 | $17.50 | 65%off |
| Claude Opus 5claude-opus-5 | $1.75 | $2.1875 | $0.175 | $8.75 | 65%off |
| Claude Opus 4.8claude-opus-4-8 | $2.40 | $3.00 | $0.24 | $12.00 | 52%off |
| Claude Opus 4.7claude-opus-4-7 | $1.78 | $2.225 | $0.178 | $8.93 | 64%off |
| Claude Opus 4.6claude-opus-4-6 | $2.40 | $3.00 | $0.24 | $12.00 | 52%off |
| Claude Haiku 4.5claude-haiku-4-5 | $0.54 | $0.675 | $0.054 | $2.70 | 46%off |
Prices in US dollars per 1M tokens, live from the catalog. Cache writes are the 5-minute kind Claude Code uses. Anthropic's list prices for cache reads and writes are its input price times 0.1 and 1.25.
Calculator
What a month costs. Your numbers.
Start from a coding-agent workload and change any number: requests, tokens in and out, and how much of the prompt Claude Code reads from the cache.
An agent like Claude Code: big prompts, mostly read from the cache.
10 models, cheapest first
- Claude Haiku 5.5Anthropic60%off$3.38
$8.45list - Claude Haiku 4.5Anthropic46%off$45.63
$84.50list - Claude Sonnet 5.5Anthropic55%off$76.05
$169.00list - Claude Opus 5.5Anthropic65%off$118.30
$338.00list - Claude Opus 5Anthropic65%off$147.88
$422.50list - Claude Opus 4.7Anthropic64%off$150.64
$422.50list - Claude Opus 4.8Anthropic52%off$202.80
$422.50list - Claude Opus 4.6Anthropic52%off$202.80
$422.50list - Claude Fable 5.1Anthropic65%off$295.75
$845.00list - Claude Fable 5Anthropic65%off$295.75
$845.00list
Setup
Two minutes. Five variables.
- Set
ANTHROPIC_BASE_URLtohttps://api.zurelay.com, with no/v1. Claude Code adds/v1/messagesitself. - Set
ANTHROPIC_AUTH_TOKENto your Zurelay key. It's sent as a Bearer token, with no approval prompt. - Point the Opus, Sonnet and Haiku aliases at Zurelay models, then run
claude. - To keep the setup, put the same values in the
envblock of~/.claude/settings.json. In the VS Code extension, add them underclaudeCode.environmentVariables.
The Claude Code guide has the shell version too. Claude Code runs on Claude models; for GPT, Gemini or DeepSeek in a coding agent, use one of the tools below.
export ANTHROPIC_BASE_URL="https://api.zurelay.com"export ANTHROPIC_AUTH_TOKEN="$ZURELAY_API_KEY"export ANTHROPIC_DEFAULT_OPUS_MODEL="claude-opus-5-5"export ANTHROPIC_DEFAULT_SONNET_MODEL="claude-sonnet-5-5"export ANTHROPIC_DEFAULT_HAIKU_MODEL="claude-haiku-5-5"claude --model claude-opus-5-5Does Claude Code work with Zurelay?
Yes. Claude Code speaks Anthropic's Messages API, and Zurelay serves Claude through that same API. Set ANTHROPIC_BASE_URL to https://api.zurelay.com (no /v1) and ANTHROPIC_AUTH_TOKEN to your Zurelay key, point the Opus, Sonnet and Haiku aliases at Zurelay model IDs, then run claude. The same variables work in the VS Code extension.
Do I need an Anthropic account or a Claude subscription?
No. You pay Zurelay per token from prepaid credit, from $10, with no subscription or minimum spend. Your Zurelay key is the only credential Claude Code needs.
How much cheaper is it?
Every Claude model costs 46–65% less than at Anthropic's API list price, on every kind of token: input, cache writes, cache reads and output. So a session that would cost $100 on Anthropic's API at list price costs about $35.00 on Claude Opus 5.5.
Does prompt caching work?
Yes. Claude Code's cache_control marks pass through as Anthropic documents them, and cached tokens are billed at Zurelay's cache prices: cache reads at a tenth of the input price, and 5-minute cache writes at 1.25 times it, as Anthropic bills them.
Which models should Claude Code use?
Map the Opus alias to Claude Opus 5.5 for the main work and the Haiku alias to Claude Haiku 5.5: Claude Code uses Haiku for background tasks, and it's the lowest-priced Claude model. Thinking levels, including xhigh and max, work as they do with Anthropic.
Is it the real Claude?
Yes. Each model is served over several routes run by independent sellers. Every route is tested before it serves requests and every day after, with identity, dated-knowledge and reasoning checks that catch a different or older model sold under the name you asked for. Failed attempts move to another route and aren't billed.
Can my team share one balance?
Yes. Credit belongs to the workspace and every key in it draws from it. Give each person or machine its own key, with its own monthly or daily budget and rate limit.
More pricing
- Claude API pricing
- GPT API pricing
- Gemini API pricing
- DeepSeek API pricing
- Grok API pricing
- Qwen API pricing
- GLM API pricing
- MiniMax API pricing
- MiMo API pricing
- Nano Banana API pricing
- GPT Image API pricing
- AI image generation API
- Seedance API pricing
- AI video generation API
- Text-to-speech API pricing
- LLM API cost calculator
- Anthropic API alternative
- OpenAI API alternative
- fal.ai alternative for video
- Claude vs GPT API pricing
- Claude vs Gemini API pricing
- GPT vs Gemini API pricing
- DeepSeek vs Claude API pricing
- OpenRouter alternative
- Every model and price
Stop paying list price.
Start saving today.
Create a key in seconds, point the SDK you already use at Zurelay, and every request costs up to 90% less from the first token.