Nomvel
Start a project
All products

Product 02 · apmix.ai

Live at apmix.ai

APMix

One key and one monthly bill for every AI model that matters.

APMix puts GPT, Claude, Gemini, Grok, DeepSeek, Qwen and more behind a single API key and a single monthly plan. Point Claude Code, Codex CLI, Cursor or your own code at api.apmix.ai, then switch models by changing one string in the request.

Start for freeSee pricing

Point Claude Code at it

ANTHROPIC_BASE_URL=https://api.apmix.ai
ANTHROPIC_API_KEY=apx_live_••••••••••••

Same key, same plan, every model. Codex and other OpenAI-compatible clients use https://api.apmix.ai/v1 instead — change the base URL and nothing else.

Status
Generally available
Category
AI model gateway
Model
Subscription
Website
apmix.ai

01Pricing

Pay once, use them all.

Priced per month in US dollars against a single shared token allowance — fast models draw on it at face value, frontier and reasoning models draw more. There are no daily or weekly caps, and every plan gets the same keys, the same endpoints and the same dashboard.

Starter

For side projects and trying things out.

$4.99/mo

Tokens
60M / mo
Models
14 included
API keys
2
Choose Starter

Pro

For daily coding with CLI agents.

$14.99/mo

Tokens
200M / mo
Models
29 included
API keys
10
Choose Pro

Max

Most popular

For heavy, all-day agent workloads.

$29.99/mo

Tokens
500M / mo
Models
41 included
API keys
Unlimited
Choose Max

Every plan includes

  • One key and one allowance across every model
  • Anthropic- and OpenAI-compatible endpoints
  • Claude Code, Codex CLI, Cursor and the official SDKs
  • No daily or weekly rate limits
  • One-time top-ups that never expire
  • 3M free tokens to try it with

01bBusiness

Built for teams and scale.

APMix Business gives your team centralized billing, usage analytics, and team management — all while keeping the same unified API access to every model.

Team management

Add team members, assign roles, and manage permissions from a centralized dashboard. Each member gets their own API keys with usage tracked individually.

Centralized billing

One invoice for the entire team with detailed breakdowns by member and project. Support for purchase orders, annual contracts, and custom payment terms.

Usage analytics

Deep insights into token consumption, model usage patterns, and cost allocation across teams and projects. Export reports for accounting and budgeting.

Priority support

Dedicated support channel with faster response times, implementation guidance, and direct access to engineering for critical issues.

Custom allowances

Flexible token allowances tailored to your team's needs. Scale up or down monthly, with high-volume discounts and predictable pricing.

SSO & security

Single sign-on integration with your identity provider, audit logs for compliance, and IP allowlisting for enterprise security requirements.

Explore Business

02Coverage

Every model behind one key.

Eleven model families, one endpoint and one invoice. When a provider is added it appears across the keys you already hold, with no migration on your side.

  1. 01

    Every frontier family

    OpenAI GPT, Anthropic Claude, Google Gemini, xAI Grok, DeepSeek, Alibaba Qwen, Z.ai GLM, Moonshot Kimi, MiniMax, Xiaomi MiMo and Tencent Hunyuan.

  2. 02

    Two compatible endpoints

    An Anthropic-compatible base URL for Claude Code, and an OpenAI-compatible /v1 endpoint for Codex, the official SDKs and curl.

  3. 03

    Switch models with one string

    The key and the base URL never change. What changes is the model named in the request.

  4. 04

    Built for agent sessions

    Cache reads are weighted at a fraction of face value, so a long conversation that reuses its context keeps costing less as it goes on.

  5. 05

    Keys you control

    Two keys on Starter, ten on Pro, unlimited on Max — and every one of them revocable from the dashboard.

  6. 06

    Usage on one meter

    Raw input, output, cache-read and cache-write counts beside the weighted total, covering every key on the account.

03Billing

The maths is on the invoice.

Weighted tokens are the whole pricing model, so it is worth being precise about. The raw counts sit next to the weighted total in the dashboard, and the weighted view can be switched off.

Cache reads cost less
Cache read tokens are weighted well below their face value. That discount is what makes an all-day agent session affordable at all.
Reasoning costs more
Reasoning-mode output draws more from the allowance than standard output does. It is the single rule most likely to move a bill.
Fast and frontier are not equal
A frontier or reasoning model spends your allowance faster than a fast one, so a mixed month costs more than the same volume on cheap models.
Top-ups, not overage
When the allowance runs out, requests pause rather than billing you at a surprise rate. Top-ups come in 10M, 30M, 60M and 100M sizes, from $1.49, and never expire.
Nothing to unwind
Cancellation takes effect at the end of the paid month, and any top-up you have already bought stays on the account.

04Availability

Open now

Every model, one key.

APMix is generally available at apmix.ai. Starter starts at $4.99 a month, and every new account gets 3M free tokens — enough to point a CLI at it and watch it work before you spend anything.

Open apmix.aisupport@apmix.ai
Website
apmix.ai
Business
business.apmix.ai
API
https://api.apmix.ai
Support
support@apmix.ai
Operated by
Zyloo, LLC