Migrate in two lines
Keep the OpenAI request and response shapes. openai-python, openai-node, and any SDK, framework, or agent client that speaks the OpenAI protocol can point here directly — your existing code stays untouched.
OpenAI-compatible · Pay per token · No minimum spend
$1 buys $10 in credit, billed per token. One base_url, 48+ models.
One-click Google / GitHub sign-in · No card required · $1 credit on sign-up ≈ 10M input tokens at our lowest per-token price
- Top-up to credit ratio
- 1 : 10
- Models to call
- 48+
- OpenAI-compatible API
- Drop-in
- Request & response content
- Never stored
https://api.y-api.bestvirtualgoods.com/v1
from openai import OpenAIclient = OpenAI( base_url="https://api.y-api.bestvirtualgoods.com/v1", api_key="YOUR_API_KEY",)response = client.chat.completions.create( model="deepseek/deepseek-v4-flash", messages=[{"role": "user", "content": "Hello!"}],)Request log
tencent/hy3↑14.7K ↓2.3KFree — no credit deductedqwen/qwen3.8-max↑12.5K ↓1.5K-$0.034anthropic/claude-opus-4.6↑2K ↓400-$0.02openai/gpt-5.5-pro↑500 ↓200-$0.051
Sample rows; your console itemizes every call like this.
Free models
Models that cost nothing per call
Every call to these models deducts nothing from your balance. Same base_url, same key, same SDK — the model field is the only thing you change, so they are also the fastest way to try Y-API before you top up.
Free status is set by our upstream provider and can be withdrawn — when it is, these calls start billing at the listed rate and your code does not change. The models page always shows the current state.
deepseek/deepseek-v4-flash$0minimax/minimax-m2.7$0tencent/hy3$0xiaomi/mimo-v2.5$0
Why Y-API
Solve cost and migration at the same time
Request content never stored
The prompts you send and the content the model returns are never saved. Request logs record only time, model, token counts, and credit deducted — for reconciliation, nothing more.
Top up $1, get $10 in credit
The top-up rate is 1:10. Credit is deducted for actual token consumption, and any top-up amount converts at the same rate. From $10 up, each top-up also adds the same amount again as a bonus. You pay for what you call, and nothing when you don't.
One balance, many keys
Keys under the same account share a single balance. Manage them per project or environment, and create or revoke any of them at any time without affecting the rest.
Every call accounted for
The console request log lists the time, model, token counts, credit used, and latency of every call; the overview page aggregates the last 30 days of usage by model.
No passwords stored
Only Google and GitHub sign-in are supported. The site stores no passwords, so there is no attack surface for password leaks or reset flows.
Migration cost
OpenAI-compatible API, two lines to migrate
If you already use the official OpenAI SDK, swapping base_url and api_key is the whole migration. Set model to any ID from the catalog.
Protocol reference: OpenAI Chat Completions API
from openai import OpenAI client = OpenAI(- base_url="https://api.openai.com/v1",- api_key=os.environ["OPENAI_API_KEY"],+ base_url="https://api.y-api.bestvirtualgoods.com/v1",+ api_key=os.environ["YAPI_KEY"], )Three steps
From sign-in to first call
01
Sign in
Sign in with Google or GitHub. Your account opens automatically with $1 in credit — no forms to fill out.
02
Get a key
A default key is created when your account opens — view it on the "API Keys" page in the console, or create more per project.
03
Change base_url
Point your SDK's base_url at Y-API, set model to any ID from the catalog, and start calling.
Available models
48+ LLM models, one API
Explore these featured models for token prices, model IDs and API examples. The full catalog includes every available model.
| Model ID | Input / 1M tokens | Output / 1M tokens |
|---|---|---|
| Vendor: DeepSeek | ||
deepseek/deepseek-v4-flash | Free — no credit deducted | |
| Vendor: Qwen | ||
qwen/qwen3.8-max | $2.00 | $6.00 |
| Vendor: MiniMax | ||
minimax/minimax-m2.7 | Free — no credit deducted | |
| Vendor: Z.ai | ||
z-ai/glm-5.1 | $1.40 | $4.40 |
| Vendor: Moonshot AI | ||
moonshotai/kimi-k3 | $3.00 | $15.00 |
| Vendor: Anthropic | ||
anthropic/claude-opus-4.6 | $5.00 | $25.00 |
| Vendor: Google | ||
google/gemini-4-argon | $4.00 | $20.00 |
| Vendor: OpenAI | ||
openai/gpt-5.5-pro | $30.00 | $180.00 |
| Vendor: xAI | ||
x-ai/grok-4.6 | $4.00 | $12.00 |
| Vendor: Other | ||
tencent/hy3 | Free — no credit deducted | |
Catalog updates
Fresh on the catalog
The newest listings, most recent first. Specs and prices are verified against the upstream catalog before a model goes live.
Pricing & billing
LLM API pricing, billed per token
Each model has its own input and output token rates, billed in USD credit. The request log records the exact credit deducted per call.
| You pay | Credit received | Bonus | Total credit |
|---|---|---|---|
| $5 | $50 | — | $50 |
| $10 | $100 | +$10 | $110 |
| $50 | $500 | +$50 | $550 |
| $100 | $1,000 | +$100 | $1,100 |
A top-up is not a plan tier — the conversion rate has no amount threshold, so any top-up converts at the same rate. A top-up of $10 or more additionally receives a bonus; see the terms above. Fixed 30-day plans are separate from top-ups: a plan gives you a fixed credit pool valid for 30 days, with no auto-renewal.
Top-up bonus
Top up $10 or more, get the same amount again
Every top-up of $10 or more adds an equal amount of credit on top of the standard 1:10 rate. It applies to every qualifying top-up — not just your first one.
- Credits are added within a minute of payment; the bonus arrives with them.
- If a top-up is refunded, the bonus granted for it is deducted as well.
- The bonus is credit, not cash: it is spent on API calls like any other credit and cannot be withdrawn.
How it adds up
$10 → $100 credit + $10 bonus = $110
$50 → $500 credit + $50 bonus = $550
$100 → $1,000 credit + $100 bonus = $1,100
Who it fits
Where the difference is biggest
Side projects & solo products
A demo with no revenue yet shouldn't carry fixed costs. Pay for what you use; pause the project and pay nothing.
Agents & batch jobs
Multi-round tool calls, long context, and batch processing multiply token consumption — and the unit-price gap multiplies straight into the bill.
Internal team tools
Hand out different keys per environment sharing one balance, and reconcile who called which model, when, from the logs.
Migrating off the official API
Your code is already built on the OpenAI SDK, and you don't want to rewrite the call layer just to switch providers.
Up front
A few things worth stating clearly
A model report you can inspect.
The GLM 5.3 and GPT 6 Astra tests are public, including the endpoints, results and original sources.
"$1 becomes $10" is credit, not a discount coupon
You pay $1 and your account gains $10 in credit. Credit burns at each model's multiplier, which varies widely — so it is not "official pricing at 10% off". Treat the actual deductions in the request log as the real cost.
Your request content is not stored
The prompts you send and the content the model returns pass through in memory and are never written to disk — not by us, and not for training. What we keep is call metadata only: time, model, token counts, latency, and credit deducted, for billing reconciliation.
Keys can be revoked anytime
A key's plaintext is returned on demand only when you click "Reveal"; it never sits exposed in lists. Once deleted, calls using that key fail immediately while other keys are unaffected.
No minimum spend
Credit is deducted only for calls you actually make. An idle account costs nothing — there is no plan to renew and no floor under the balance, and the $1 sign-up credit arrives without a card.
FAQ
You might also wonder
- What is Y-API?
- Y-API is an OpenAI-compatible LLM API service. It provides a single endpoint and key so the same code can call 48+ models — including DeepSeek, Qwen, MiniMax, Z.ai, Moonshot AI, Anthropic, Google, OpenAI, xAI, Tencent, Xiaomi, and StepFun — billed per token.
- Is Y-API compatible with the official OpenAI API?
- Yes. Y-API uses the same request and response structure as OpenAI. Swap the SDK's base_url for Y-API's endpoint and the api_key for a Y-API key — nothing else changes.
- Which models does Y-API support?
- Y-API currently supports 48+ models from vendors including DeepSeek, Qwen, MiniMax, Z.ai, Moonshot AI, Anthropic, Google, OpenAI, xAI, Tencent, Xiaomi, and StepFun. The full list of model IDs is on the Y-API models page (https://y-api.bestvirtualgoods.com/models); pass any of them in the request's model field.
- Do any models cost nothing at all?
- Yes — deepseek/deepseek-v4-flash and minimax/minimax-m2.7 and tencent/hy3 and xiaomi/mimo-v2.5 are free through Y-API: no credit is deducted per call, using the same base_url and key as the paid models. The free flag is set by our upstream provider and can be withdrawn; if it is, these calls start billing at the listed rate, with no code change on your side.
- How does Y-API billing work?
- Account credit is deducted based on token consumption. The deduction depends on the called model's multiplier, with input and output counted separately. The exact deduction for any call is in the console request log.
- What does "top up $1 and get $10 in credit" mean?
- Every $1 you top up adds $10 to your account credit. Credit is the billing unit inside the service, deducted per token; different models burn it at different rates, so it is not a fixed discount off official pricing.
- Does Y-API have a minimum spend?
- No. Y-API only deducts credit for actual token usage, and an idle account costs nothing.
- Can I create multiple API keys? Is credit split between them?
- Yes, you can create multiple keys, and they all share the same account balance — no per-key allocation. This suits per-project or per-environment management; any key can be revoked at any time without affecting the others.
- Does Y-API support streaming?
- Yes. Add the stream parameter to the request and you get SSE chunks in the same format as OpenAI — no special client handling needed.
- Can I see my call history?
- Yes. The console request log lists the time, model, key, token counts, credit used, and latency of every call, and the overview page aggregates the last 30 days of usage and request counts by model.
- Can Y-API see the content I send?
- No. Y-API does not store request or response content. The logs shown in the console contain only call metadata (time, model, token counts, latency, credit deducted) — never prompts or completions.
- What happens when my credit runs out?
- The API returns HTTP 403 with error code insufficient_user_quota — not 429 — and the request is not executed. Top up at https://y-api.bestvirtualgoods.com/app/billing — credit is available immediately, keys stay valid, and no code changes are needed.
- Do I need a credit card to sign up?
- No. Sign in with Google or GitHub and your account opens automatically with $1 in credit — no payment details involved during the trial.
- Are there rate limits for trial accounts?
- After sign-up, you are limited to 100 requests per day (shared across all models, regardless of whether they are free or paid). Checking in once a day raises that day's allowance to 200. A single top-up of $5 or more lifts the limit permanently; paid models are additionally subject to account balance constraints, with credit deducted based on token usage.
Start now
Sign in and your account opens automatically with a default key and $1 in credit. No card, no forms.