01
Create a key
Generate one sk-pool-* key. Every Claude, GPT and Gemini model runs on a single shared balance.
- All models
- Claude, GPT & Gemini
- One balance
Claude + GPT · One gateway
Buy Claude, GPT & Gemini API accessSame as official, but cheaper
- Integration with Claude Code, Codex, Cursor, opencode & more
- Add it to any product or workflow
- No Anthropic or OpenAI account
- No VPN
10+Models
3API formats
50%B2C discount
2Payment options
minutesSetup time
What you get
Official Claude, GPT and Gemini APIs, at a discount
Top up in plain USD and spend it at official provider rates — minus a flat 50% discount on every request. Every Claude, GPT and Gemini model, one key, one balance.
Full-time development with an AI coding agent — all day, every day, ~22 workdays.
You pay for a month of coding
$106.25
for a month of coding on Claude Fable 5.1 · official Anthropic price minus your 50% discount
You save$106.25≈ ×2 more work per $
Get API keySame Anthropic Messages API, models and responses. You just pay less per call.
Every Claude model for this task
Official Anthropic price vs your apiToken.sale price at −50%, for a month of coding.
| Model | Input / 1M | Output / 1M | Official | Your price | You save |
|---|---|---|---|---|---|
Claude Fable 5.1claude-fable-5-1 | $10 | $50 | $212.50 | $106.25 | $106.25 |
Claude Opus 5.5claude-opus-5-5 | $4 | $20 | $93.00 | $46.50 | $46.50 |
Claude Opus 5claude-opus-5 | $5 | $25 | $136.25 | $68.13 | $68.13 |
Claude Fable 5claude-fable-5 | $10 | $50 | $272.50 | $136.25 | $136.25 |
Claude Opus 4.8claude-opus-4-8 | $5 | $25 | $136.25 | $68.13 | $68.13 |
Claude Opus 4.7claude-opus-4-7 | $5 | $25 | $136.25 | $68.13 | $68.13 |
Claude Sonnet 5.5claude-sonnet-5-5 | $2 | $10 | $46.50 | $23.25 | $23.25 |
Claude Sonnet 5claude-sonnet-5 | $2 | $10 | $54.50 | $27.25 | $27.25 |
Claude Sonnet 4.6claude-sonnet-4-6 | $3 | $15 | $81.75 | $40.88 | $40.88 |
Claude Haiku 5.5Cheapestclaude-haiku-5-5 | $0.1 | $0.5 | $2.73 | $1.36 | $1.36 |
Claude Haiku 4.5claude-haiku-4-5 | $1 | $5 | $27.25 | $13.63 | $13.63 |
Each task uses a realistic total number of tokens to finish the whole job — expand “Adjust the tokens yourself” to tune it. Cache reads use 0.1× input except Claude Fable 5.1 at 0.025×; 5-minute cache writes use 1.25× input. Estimates only — your real bill depends on exact token usage.
Full calculator — prompt caching, every model & pricing FAQ →
Get started
Live in three steps
Create a key, point any compatible tool at the matching endpoint, and ship — Claude, GPT and Gemini models, streaming, lifetime spend limits and optional key expiration on a single balance.
02
Pick your API surface
One unified endpoint: https://router.apitoken.sale serves the native Anthropic, OpenAI and Gemini lanes plus an OpenAI-compatible universal lane for any catalog model. Keep the tools and SDK you already use — the legacy per-provider hosts remain supported.
- Claude Code
- Codex
- Cursor
- Cline
- opencode
- SDK
03
Ship to production
Full Messages API plus Responses and Chat Completions with SSE streaming, per-key lifetime spending limits, optional daily and monthly spend caps, optional expiration, model allowlist, request history and official-spend clarity.
- SSE streaming
- Key guardrails
- Usage & spend
Pricing
Prepaid balance — 50% off every request
One flat 50% B2C discount on official provider prices — the same rate for every request and every top-up. B2B pricing is negotiated.
Flexible top-up
Choose any whole USD amount
≈ $2,000−50%
of official API usageNo catalog and no preset amounts. Enter a whole amount and add exactly that much to your balance.
Create accountB2B pricing
Negotiated business pricing
Pricing modelCustom rate
Private accessDiscountVolume-based
OnboardingDirect with our team
Business customers receive an operator-set discount under an individual agreement.
Request B2B accessB2C · Flat pricing
One flat 50% discount on every request
Every request50% off official provider prices
Every top-upAny whole USD amount at the same rate
| Top up | Discount | Official API value |
|---|---|---|
| $50 | 50% ×2 | $100 |
| $100 | 50% ×2 | $200 |
| $1,000 | 50% ×2 | $2,000 |
$5050% ×2
Discount−50%
Official API value$100
$10050% ×2
Discount−50%
Official API value$200
$1,00050% ×2
Discount−50%
Official API value$2,000
Official API value = top-up ÷ the share paid after the 50% discount. Values are rounded only for display; billing remains exact.
Estimate your cost in the free calculator →FAQ
Common questions
Short answers to the questions that usually matter before buying an API key.
How much does API access cost?
Enter any positive whole USD top-up amount. Each request is converted to official API spend, then your active discount is applied: B2C takes a flat 50% off every request, while B2B pricing is negotiated.
How do I start using the API?
Create an account, top up your balance, and generate an API key. API usage is charged from your prepaid balance.
Which models are available?
The Claude line — Opus 5.5, Opus 5, Fable 5, Opus 4.8, Opus 4.7, Sonnet 5.5, Sonnet 5, Sonnet 4.6, Haiku 5.5 and Haiku 4.5 — plus the GPT line — GPT-6.1 Sol, GPT-6 Astra, Sol and Luna, and GPT-5.6 Sol, Terra and Luna — and the Gemini line — Gemini 3.8 Flash, 3.7 Flash, 3.6 Flash, 3 Flash Preview, 3.1 Pro Preview and more. One balance and one API key cover every model.
Which model is best for coding?
For agentic coding and long sessions, Claude Opus and Sonnet and GPT-5.6 Sol give the best results; Claude Haiku and GPT-5.6 Luna are ideal for fast, cheap calls.
What base URL should I use?
Use the unified router at https://router.apitoken.sale. Native lanes: POST /v1/messages (x-api-key) for Claude and Anthropic-compatible tools, POST /v1/responses for GPT (Authorization: Bearer), and /v1beta/models/{model}:generateContent for Gemini (x-goog-api-key). The OpenAI-compatible universal lane POST /v1/chat/completions reaches any catalog model, and GET /v1/models lists the unified catalog. Legacy per-provider hosts (api.apitoken.sale, openai.api.apitoken.sale/v1, gemini.api.apitoken.sale) remain supported for existing integrations.
What happens if I receive a 429?
Honor Retry-After and reduce concurrency. Dashboard key guardrails are a lifetime spending limit, optional daily and monthly spend caps, an optional expiration date and a model allowlist; they do not configure request throughput.