Enterprise-grade Claude API built for Claude Code
A proprietary account pool of official Claude Max subscriptions, connected directly to Anthropic. Native protocol support — set two environment variables and get stable, full-capability Claude models with no rate limits.
- hubProprietary account pool
- workspace_premiumOfficial Max subscriptions
- routeDirect to Anthropic
- verified_userEnterprise-grade service
Why choose Zivv for Claude
We focus on Claude and own every layer: the pool, the accounts, the route and the service.
Proprietary account pool
Our scheduler is built in-house: it pins each session to one account to keep Prompt Cache hits, watches every account's health and quota in real time, and fails over automatically so long sessions keep running.
Official Max subscriptions
The pool is made of official Anthropic Claude Max subscription accounts (Max 5x and Max 20x). Full-capability models across Opus, Sonnet and Haiku.
Direct to Anthropic
Requests go straight to Anthropic's official API with no third-party resellers in between. Native Anthropic protocol with extended thinking, tool use, subagents and MCP.
Enterprise-grade service
Hand out team keys, set daily / monthly / total budgets, and break usage down by member, key and model. Ticket and community support, plus a public 24/7 status page.
No rate limits
No RPM / TPM caps — requests are spread across many accounts, so high concurrency and long-context tasks never queue.
Predictable cost
Pay per token, settled at ¥1 = $1. No monthly fee, no minimum spend, no hidden costs.
Real Claude Code Workflows
Everyday coding and debugging
Let Claude Code read your repository, explain code, fix errors and run tests. Set the endpoint and key once without changing your existing commands or workflow.
Large refactors and long tasks
Handle cross-file changes, dependency work and repeated test-fix loops. The native Anthropic protocol preserves tool use, subagents and extended thinking.
Team development
Assign separate keys to each member or project, then manage spend through team budgets and usage analytics while keeping ownership clear.
Codebase and long-document analysis
Work across large codebases, technical documents and long-context tasks while reviewing spend by actual token usage to choose the right model.
Zivv vs. Official API vs. Other Relays
| Feature | Zivv | Claude Official | Other Relays |
|---|---|---|---|
| Account source | Official Max subscriptions | Official API | Often undisclosed |
| Upstream route | Direct to Anthropic | Direct | Often multi-hop resale |
| Pricing | 1.2× official price, settled at ¥1 = $1 | $3–15 / M tokens | Varies by provider |
| Usage limits | Proprietary pool scheduling | Subject to official quotas | Varies by provider |
| Protocol compatibility | Native Anthropic protocol | Official Anthropic API | Varies |
| Private deployment | Available | Not offered | Rarely offered |
| Support | Tickets + community + help center | Official docs and support | Varies by provider |
Connect Claude Code in Three Steps
About 30 seconds end-to-end, no code changes.
Create a Key in the Claude MAX group
Sign in to the console, open the API Keys page, create a Key and choose "Claude MAX" as the group.
Set environment variables
Add these two lines to your shell config, replacing ANTHROPIC_AUTH_TOKEN with the Key you created.
export ANTHROPIC_BASE_URL=https://zivv.pro
export ANTHROPIC_AUTH_TOKEN=<YOUR_API_KEY>infoANTHROPIC_BASE_URL uses the root domain — no /v1 suffix required.
Launch Claude Code
Just run claude. Model selection, subagents, MCP and extended thinking all work as expected.
Pricing
Billed at 1.2× the official model price, but settled 1:1 in RMB (¥1 = $1). Compared with buying the official API at the live exchange rate, your actual RMB spend is significantly lower.
Billing Ratio
Based on official model pricing
Settlement
Top up and use, no currency conversion
Effective Official Price
Estimated at live rate ¥7.30 = $1
- check_circleFull Claude model coverage, synced on day one of each release
- check_circleNo rate limits, no queuing under high concurrency
- check_circlePay per token, nothing more
- check_circleSupports extended thinking, tool use, subagents and MCP
Private account pool deployment
Run the same proprietary account pool that powers Zivv on your own servers. Your accounts, data and traffic stay with you — ideal for companies, teams and providers with their own Claude subscriptions.
Production-grade pool
The same scheduler Zivv runs in production: session affinity, account health checks and automatic failover out of the box, kept up to date with Anthropic changes.
Fully private data
Account credentials and request logs live in your own database, and requests go from your server straight to Anthropic — never through Zivv.
Admin panel
Add accounts, check health, quota usage and request logs, and issue keys to your team from one panel.
Fast Docker setup
Two containers on a single 2-core / 2 GB server. Deploy it yourself, or let us deploy it remotely for you.
How it works
- 1Contact us about your account count and deployment needs
- 2We issue a license and grant image access
- 3Deploy it yourself, or we deploy it remotely
- 4Add your Claude accounts and connect Claude Code
infoAccount and instance limits are set per license. Trials available.
Ask about private deploymentFAQ
Get Started with the Claude MAX Group
Sign up free, select the Claude MAX group when creating a Key, and connect Claude Code in about 30 seconds.