OPENAI-COMPATIBLE API · FIXED 30-DAY QUOTA
The AI API platform with a fixed monthly price
One OpenAI-compatible API key for astra, sol-6.1, sol-6, terra and luna-6: a hard 30-day quota with no overage fees, automatic retries when a model provider is overloaded, and live usage in your portal.
Independently operated — who runs this and where is on the About page. About
Estimated equivalent cost of the 318,420,000 credits spent this cycle, at reference rates — same workload, same models.
https://api.aifromu.com/v1
What the platform does
Retries built in
When a model provider is overloaded before it starts answering, the platform asks again on your behalf, so you usually get the answer instead of the error.
Hard quotas, no overage
Every plan is a fixed 30-day credit quota. When it runs out, requests stop: no surprise bill.
Live usage and cost
Your portal shows credits next to raw tokens for every request, by model and by hour.
A request monitor
See every request's status, timing and why it failed, and whether it was charged.
What you can count on
The real model, never swapped
Every answer is checked against the model you asked for. If any other model replies, the gateway stops that answer and you are not charged for it.
No middleman
Your request goes from our gateway straight to the model provider, never through anyone else.
Your code is never stored
We keep token counts for your bill, never your prompts, your code or the answers, and none of it is used to train models. See our Privacy policy.
// 01 · PRICING
Four plans. Zero overage.
Starter
Growth
Pro
Scale
When credits reach zero, new requests stop. Your card is never charged for overage.
// 02 · SAVINGS
More model for the same budget.
ESTIMATEDIllustrative cost per 1M short-context uncached input tokens
| MODEL | INPUT | OUTPUT | DIFFERENCE | ||
|---|---|---|---|---|---|
| REFERENCE RATE | AI FROM U SCALE, EST. | REFERENCE RATE | AI FROM U SCALE, EST. | ||
| astra | $10.00 | $0.30 | $50.00 | $1.5 | ≈ −97.0% |
| sol-6.1 | $2.00 | $0.060 | $10.00 | $0.30 | ≈ −97.0% |
| terra | $1.00 | $0.030 | $5.00 | $0.15 | ≈ −97.0% |
Same $60. Up to 33.3× the short-context uncached input tokens.
Estimated equivalent cost on the Scale plan, input-token basis. Your effective rate depends on how you use your quota — the point stands: the same budget buys you more model.
How this is estimated
- Holds when the full 1,100,000,000 credits of the Scale plan ($60) are used within the 30-day window.
- Cache reads convert at half the model's input rate and cache writes at the input rate itself; output converts at 5× the input rate.
- Comparisons apply once a model is bound to its exact upstream counterpart.
- Reference rates are indicative and not a verified provider quote. We will label them as provider list price only once verified and dated.
// 03 · CREDITS
Credits are your quota. Raw tokens are the receipt.
AI credits are AI From U's quota unit. Each request converts provider-reported usage into credits using model-specific rates.
Your portal shows credits alongside raw-token usage, so you can verify what your workload consumed.
Deducted from your plan. When the balance reaches zero, new requests stop.
Provider-reported input and output usage, including reasoning and cache breakdowns.
A token read from the cache is charged at half its model's input rate — 2.75× on Astra, 0.55× on Sol 6.1 and Sol 6, 0.275× on Terra, 0.0275× on Luna 6 — and your portal shows that discount as its own line on every request.
// 04 · MODELS
Five models. One endpoint.
sol and luna still work as model names for existing integrations: they now run sol-6.1 and luna-6, billed at those models' rates. Requests that name GPT-5.6 Terra are served by terra, which runs GPT-6.1.
Photographs: NASA · Star: ESA/Hubble & NASA
Weights affect credits charged, not raw-token counts.
How AI credits are calculated
One input token on Sol 6.1 or Sol 6 is 1.1 AI credits. Those two are the anchor, and 1.1 is their factor. Every model has one factor: input and cache writes count it whole, cache reads half of it, and output and reasoning 5× it, the ratio the provider's own list price sets. Astra's factor is 5.5, Terra's 0.55 and Luna 6's 0.055. Each rate is charged exactly as published and itemised on every request in your portal.
Short context
| MODEL | INPUT | CACHE READ | CACHE WRITE | OUTPUT + REASONING |
|---|---|---|---|---|
| ASTRA | 5.5 | 2.75 | 5.5 | 27.5 |
| SOL 6.1 | 1.1 | 0.55 | 1.1 | 5.5 |
| SOL 6 | 1.1 | 0.55 | 1.1 | 5.5 |
| TERRA | 0.55 | 0.275 | 0.55 | 2.75 |
| LUNA 6 | 0.055 | 0.0275 | 0.055 | 0.275 |
Long context
| MODEL | INPUT | CACHE READ | CACHE WRITE | OUTPUT + REASONING |
|---|---|---|---|---|
| ASTRA | 5.5 | 2.75 | 5.5 | 27.5 |
| SOL 6.1 | 1.1 | 0.55 | 1.1 | 5.5 |
| SOL 6 | 1.1 | 0.55 | 1.1 | 5.5 |
| TERRA | 0.55 | 0.275 | 0.55 | 2.75 |
| LUNA 6 | 0.055 | 0.0275 | 0.055 | 0.275 |
Requests above 272K actual input tokens use long-context rates. Unclassified usage is charged using the higher applicable input or output weight.
// 05 · HOW IT WORKS
From checkout to first call
Pick the 30-day plan that matches your usage.
Complete your purchase on PayPal's secure page. AI From U is the seller; PayPal processes the payment.
Your plan activates after PayPal confirms your payment.
Receive your API key, store it securely, and start calling
https://api.aifromu.com/v1
// 06 · API
Two lines to switch
Anything that speaks the OpenAI API format works. Change the base URL, keep your code.
curl https://api.aifromu.com/v1/models \ -H "Authorization: Bearer $AIFU_API_KEY"
from openai import OpenAI client = OpenAI( base_url="https://api.aifromu.com/v1", api_key="YOUR_AIFU_API_KEY", ) resp = client.chat.completions.create( model="terra", messages=[{"role": "user", "content": "Hello from AI From U"}], ) print(resp.choices[0].message.content)
Built to be verifiable
05 / 05 GUARANTEESRequests stop when your credit balance reaches zero. No silent overage.
Only confirmed payment activates your plan. The return page never grants access.
API keys are shown once and can be rotated securely from your portal.
Payments, refunds and automatic renewals go through PayPal. AI From U is the seller, and card details stay with PayPal.
Credits are shown beside provider-reported raw-token usage, so you can verify every deduction.
// 07 · Blog
From the blog
What we learn running an AI API platform: how a request is priced, what the meter records, and which model to pick for the job.
-
GPT-6.1 Terra is here: which model should you use?
GPT-6.1 Terra costs half what Sol does. What Terra, Sol and Astra each cost, where each wins on OpenAI's own benchmarks, and which one to pick for each task.
-
Choosing between astra, sol, terra and luna
A procedure, not a table: start on terra, move a class of request to sol when reasoning pays, spend astra deliberately, and leave the background to luna.
-
Using Cursor with your own base URL
Cursor speaks the OpenAI API, so it can speak here instead: the two fields to fill in its model settings, the five models, and a command that checks both.
-
LangChain on an OpenAI-compatible API
The two fields ChatOpenAI needs to reach this endpoint, why invoke, stream and tool calls pass through unchanged, and how cache reads keep an agent loop lean.
-
Point the OpenAI SDK at a new base URL
Two changes move an existing OpenAI-SDK app here: the base URL and the key. Chat completions, responses, streaming and tool calling all arrive unchanged.
-
Streaming, and the 100-second edge
Why a long non-streaming request dies at the edge after roughly 100 seconds, how one field prevents it, and what the meter does around a stream.
// FAQ
Questions, answered
What are AI credits and raw tokens?
AI credits are AI From U's quota unit, deducted from your plan. Raw tokens are the provider-reported usage each request consumed. Your portal shows both, so you can verify every deduction.
How does cache-read pricing work?
A token read from the cache is charged at half its model's input rate — 2.75× on Astra, 0.55× on Sol 6.1 and Sol 6, 0.275× on Terra, 0.0275× on Luna 6 — a cache write at the input rate itself, and output and reasoning at 5× it. Your portal itemises the cache discount on every request, so the number you are charged is the number you can check.
What happens when I run out of credits?
New requests stop when your credit balance reaches zero. AI From U does not charge usage-based overage fees. Your subscription continues to renew according to its billing cycle.
When do I receive my API key?
After payment confirmation activates your plan. The key is revealed once in the portal — store it safely. You can rotate it at any time.
How do I manage or cancel my subscription?
In your AI From U portal, under Billing: turn off auto-renew, or pay for 30 more days on the same key. A subscription renews every 30 days until you turn it off; a plan bought once does not renew. Buying 30 more days starts them at once; credits you have left stay usable until the end of their own period.
How can I verify my usage?
Your portal shows AI credits alongside provider-reported raw-token usage, including input, output, reasoning and cache breakdowns. Recent request IDs are redacted but remain available for reconciliation.
How are my API key and payment details protected?
Your complete API key is shown only once and can be rotated from the portal. Card details are handled on PayPal-hosted checkout and billing pages; AI From U does not collect or store raw card information. We keep token counts for your bill, never your prompts, your code or the answers, and none of it is used to train models.
Is AI From U an official OpenAI service?
No. AI From U is an independent service that provides access to OpenAI models through an OpenAI-compatible API format. OpenAI provides the underlying models, while AI From U manages access, credits, usage reporting and billing. Your request goes from our gateway straight to the model provider, never through anyone else.
What do the numbers after a name mean, and what runs on sol and luna?
The number after a name is the GPT generation it runs on: sol-6.1 runs GPT-6.1, and sol-6 and luna-6 run GPT-6. The names without a number: terra runs GPT-6.1 and astra GPT-6. sol and luna still work as model names for existing integrations: they now run sol-6.1 and luna-6, billed at those models' rates. Requests that name GPT-5.6 Terra are served by terra, which runs GPT-6.1. Which model should I use?
How do I know the answer comes from the model I asked for?
Every answer is checked against the model you asked for. If any other model replies, the gateway stops that answer and you are not charged for it.