GPT · Claude · Gemini · Grok · DeepSeek — 100+ models, one key

Nobody cheaper runs this stable.
Nobody this stable runs this cheap.

One OpenAI-compatible endpoint for every major model — official list prices, members save up to 50%.
Official-grade models, enterprise stability, zero data retention. No relay games.

Drop-in: change the base URL, paste your key, pick any model.

python 
from openai import OpenAI

client = OpenAI(
    base_url="https://clssai.com/v1",
    api_key="sk-..."   # api.clssai.com/key
)

client.chat.completions.create(
    model="claude-fable-5",
    messages=[{"role": "user", "content": "Hi"}],
)
Get your API Key See models & prices

Top up and go · Official-source models · Pay per token · We never store your conversations

Stop juggling vendors, accounts and bills

One balance, one key, one OpenAI-compatible endpoint — for everything.

Going direct to each vendor
  • A separate account, key and invoice for every provider
  • Region locks, card rejections, prepaid deposits
  • Full list price — no leverage as a small team
  • Rewrite your client for each vendor's SDK
freem.ai unified gateway
  • One key and one balance for 100+ models
  • Top up in seconds — no vendor approval, no region wall
  • Roughly half the official price, with real stability
  • OpenAI-compatible — switch with one line of code

Cheaper than official, steadier than the cheap channels

The bargain relays cut corners. We don't — we just have enterprise-scale leverage.

Feature
Official direct
Bargain channels
freem.ai
Price
Full list price
Dirt cheap, until it breaks
~50% off, and stable
Models
One vendor only
Mixed, sometimes faked / nerfed
All majors, official-source
Stability
High, but region-gated
Frequent 5xx / throttling
Multi-node failover, 99.9%
Data privacy
Bound by their terms
May log / leak your data
Zero retention · DPA on request
Billing
Pay each vendor separately
Opaque, no usage view
One balance, transparent usage
Integration
A different SDK each
Non-standard endpoints
OpenAI-compatible · one line

Every major model — official source, half price

100+ models behind one key. No swaps, no nerfs — the real thing.

Official list prices — switch vendors above

USD per 1M tokens. Starter pays exactly these prices; Plus pays 90%, Pro 70%, Enterprise 50%.

Full price list for every model

Subscribe for a discount — or just pay as you go

Every membership fee comes back in full as usage credit. Need more? Top up at your tier's discount.

Starter

List price

No subscription. $1 trial credit on signup, then top up from $10 — billed at official list prices.

  • $1 free credit to try every model
  • All 100+ models at official prices
  • Balance never expires
  • Real-time usage dashboard
Start free

Plus

10% off usage
$25 / month

Includes $25 usage credit every month — the fee comes straight back.

  • $25 credit / month included
  • All usage at 90% of list price
  • Top-ups billed at the same 10% off
  • Priority routing nodes
Most Popular

Pro

30% off usage
$250 / month

Includes $250 usage credit every month, with every token at 30% off.

  • $250 credit / month included
  • All usage at 70% of list price
  • Top-ups billed at the same 30% off
  • Higher rate limits & priority nodes

Enterprise

50% off usage
$2,500 / month

Includes $2,500 usage credit every month — the true half-price tier, built for scale.

  • $2,500 credit / month included
  • Half the official price on everything
  • Dedicated nodes, SLA, invoices & DPA
  • Named account manager

Pay 6 months upfront for 10% off the fee, or a year upfront for 20% off. Usage discounts apply to every model and every top-up. Exact rates: see the full price list.

Plug in anywhere

If it speaks OpenAI, it already speaks freem.ai. More clients land every week.

Live

OpenAI-compatible API

Point any OpenAI SDK or HTTP client at our base URL. Chat, embeddings, vision, tools.

Get a key & docs
Live

Claude Code

Run Claude Code on freem.ai compute — same workflow, half the cost. Step-by-step guide.

Claude Code setup guide
Beta

Claude Desktop App

Claude desktop app (Chat · Cowork · Code) on freem.ai compute. Mirror download, no VPN.

One-click install
Live

Codex CLI

OpenAI Codex in your terminal on freem.ai compute. Tested end-to-end via the Responses API.

One-click install
Beta

Codex Desktop App

Codex desktop app — parallel threads & worktrees. Mirror download, shares CLI config.

One-click install
Live

OpenClaw 🦞

Your own AI assistant on your devices, routed through freem.ai. No subscription.

One-click install
Live

Hermes Agent

Nous Research's Hermes — memory, automation, sandbox. OpenAI-compatible, BYO key.

One-click install
Coming soon

More clients

Cherry Studio, ChatBox, Cline, NextChat and more — one-click presets on the way.

Straight answers

Are these the real models, or watered-down copies?

The real ones. Every request hits the official upstream API unmodified — no nerfing, no swaps, no fake models. You get the exact model you asked for.

Do you store or train on my data?

No. We don't retain, cache or train on your prompts or responses. Conversations pass through and are gone. Enterprise can sign a DPA.

How can it be half price and still stable?

Enterprise-scale volume plus multi-supplier redundancy and our own gateway. We pass the scale discount to you instead of cutting corners.

How do recharging and billing work?

Top up at api.clssai.com/key, then it's billed per token in real time with a transparent usage dashboard. Balance never expires.

How do I connect?

Use the OpenAI-compatible API, or follow our Claude Code guide at x.clssai.com/cc. More client presets are landing soon.

Get your API Key