Hi, I’m Reeva — your business’s AI
Reeva AI Models API Unlimited tokens. One flat monthly price.
I summarize emails, answer customers, extract data, write code and draw banners — through one OpenAI-compatible API that works in the tools you already use. Unlimited tokens, from $5 a month.
- Unlimited tokens
- 3 models
- OpenAI-compatible API
- Monthly, cancel any time
Summarize this email thread with our supplier and tell me what to do next.
Re: order 4821 · 7 messages
Summary:
- The supplier confirmed that 12 monitors ship on Monday; the other 3 follow a week later.
- The unit price dropped 8% after the volume discount.
- Open: they need written approval of the partial delivery by tomorrow.
Next step:
Reply “Approved: 12 now, 3 next week” and ask for an updated invoice.
Works with any tool that speaks the OpenAI API
- opencode
- Continue
- Aider
- Cline
- Roo Code
- OpenAI Python SDK
- OpenAI Node SDK
- curl
One subscription
Reeva brings AI to every business.
Three models, one API and a key for every tool — from the office to the laptop at home. No IT department, no separate subscription for every app.
- models
- 3
- tokens per month
- ∞
- keys per account
- 10
- requests / min on Scale
- 120
The AI gateway
Every model. One API. Your key.
Tools that already work with OpenAI work with Reeva: change the base URL and the key. A separate key for every tool, clear rate limits for every plan, and unlimited tokens at a flat monthly price.
- Chat and images on the standard OpenAI paths, plus embeddings on Pro and Scale
/v1/chat/completions/v1/images/generations/v1/images/edits - Up to 10 keys, each with its own name and optional expiry, each revocable on its own
- Extra requests wait up to two minutes for a free slot, and the OpenAI SDKs retry a 429 on their own
Three models, in every plan
All three run on Reeva’s own servers, with no third-party cloud AI, behind one OpenAI-compatible endpoint. Switch models by changing the model id.
DeepSeek v4.1
Chat, code and documents
Fast general-purpose model and the default choice for chat, summaries, drafting, translation, code and pulling data out of documents and emails. The email, invoice, code and translation examples in the demo run on it, and it works in opencode, Cline, Continue and Aider with your Reeva key. It takes the full request size your plan allows: up to 32k tokens on Mini, 128k on Starter and Pro, and 256k on Scale, so long documents and large codebases belong here.
API id: deepseek-v4.1-flash
Qwen-Image
Image generation
Generates images from a text prompt on /v1/images/generations and edits images you already have on /v1/images/edits: banners, product shots and campaign visuals. An image takes longer to come back than a text answer. Past 3 images an hour on Mini, 10 on Starter, 30 on Pro or 60 on Scale, your requests slow down.
API id: qwen-image
Qwen 3.6
Light, quick tasks
A smaller model for light, quick tasks: short replies, rewording, simple drafting and translation. Requests up to 32k tokens, prompt and answer together, on every plan; a larger request is refused with HTTP 400. Use DeepSeek v4.1 for anything longer.
API id: qwen3.6-35b-a3b
Embeddings for search and RAG
Turn text into vectors on the standard /v1/embeddings path with nvidia’s Nemotron-3-Embed-8B: embed your documents once, embed each question as it comes, and rank the documents by how close they are. The model has two ids, one for each side of the search.
nemotron-embed-8b- for the documents and passages you store
nemotron-embed-8b-query- for the questions you search with
Use both ids, each for its own side. Embedding questions and documents with the same id returns no error, but the search ranks the results much worse.
- 4096-dimension vectors, already normalized: a dot product is the cosine similarity
- Up to 64 inputs and 120,000 characters per request
- Each input is embedded up to 4,096 tokens; longer text is cut without an error, so split long documents into chunks first
- Embedding tokens count the same as chat tokens, and a request takes one of your slots for requests at once
from openai import OpenAI
client = OpenAI(base_url="https://api.reeva.me/v1",
api_key="YOUR_API_KEY")
docs = ["Invoices are due 30 days after issue.",
"Refunds go back to the original card."]
d = client.embeddings.create(
model="nemotron-embed-8b", input=docs)
q = client.embeddings.create(
model="nemotron-embed-8b-query",
input="When is an invoice due?")
# Normalized vectors: dot product = cosine similarity.
qv = q.data[0].embedding
scores = [sum(a * b for a, b in zip(qv, x.embedding))
for x in d.data]
print(docs[scores.index(max(scores))])What you do with it
Work that does itself. Starting today.
Six jobs every business does every day — each one a single API request. Click a card to watch it run in the demo.
deepseek-v4.1-flash Email thread summaries
A seven-message thread with a supplier becomes three bullets and one action item — with the numbers right.
qwen3.6-35b-a3b Customer replies
A to-the-point draft in your tone, from email or your CRM. You approve it and send it.
deepseek-v4.1-flash Data from documents
An invoice, a quote or a form becomes clean JSON for your software: supplier, number, date, amounts.
deepseek-v4.1-flash Code and agents
Completion, review and tool use — in opencode, Cline, Continue and Aider, with your key.
qwen-image Campaign images
Banners, product shots and edits — generated and edited through /v1/images.
deepseek-v4.1-flash Translation and drafting
Between languages, in a marketing or formal tone, shorter or longer. For the website, the catalog or a tender.
How to start
Plan, key, paste. A few minutes.
-
Choose a plan
Pick Mini, Starter, Pro or Scale and pay with PayPal. Your account is created and your first key is issued as soon as the payment clears.
-
Create more keys
Give each tool its own key, with a name and an optional expiry. Up to 10 keys; revoke any one of them on its own.
-
Paste it into your tool
Base URL and key — in Python, curl, opencode or Cline. The same code you wrote for OpenAI.
from openai import OpenAI
client = OpenAI(
base_url="https://api.reeva.me/v1",
api_key="YOUR_API_KEY",
)
r = client.chat.completions.create(
model="deepseek-v4.1-flash",
messages=[{"role": "user",
"content": "Summarize this email in three bullets"}],
)
print(r.choices[0].message.content)A different base_url and a different key. That’s all.
AI API plans
Unlimited tokens on every plan. You pay for speed.
No monthly token cap and no per-token billing. Plans differ by requests per minute, requests at once and maximum request size, and Pro and Scale add embeddings. Monthly, no commitment, cancel any time.
Prices in USD, billed monthly through PayPal.
Qwen 3.6 takes requests of up to 32k tokens (prompt plus answer) on every plan; use DeepSeek v4.1 for longer ones.
Enterprise
Need more? Dedicated instance, custom rate limits, SLA and contractual privacy terms.
AI Models API — frequently asked questions
Which models does the AI Models API include?
Every plan includes three models: DeepSeek v4.1 for chat, summaries, drafting, code and long documents (API id deepseek-v4.1-flash), Qwen-Image for image generation and editing (qwen-image), and Qwen 3.6 for light, quick tasks (qwen3.6-35b-a3b). Plans differ only in how hard you can use them: requests per minute, how many requests run at once, how much you can use in an hour before requests slow down, and the largest single request — 32k tokens on Mini, 128k on Starter and Pro, and 256k on Scale. Qwen 3.6 is the exception: up to 32k tokens per request on every plan. gpt-oss-20b has been retired, and requests that name it are refused: use deepseek-v4.1-flash for code and agents.
Does the API create embeddings?
Yes, on Pro and Scale; Mini and Starter do not include them. Send POST /v1/embeddings in the OpenAI format to nvidia’s Nemotron-3-Embed-8B, which has two ids, and you need both: nemotron-embed-8b for the documents you store and nemotron-embed-8b-query for the questions you search with. Using one id for both returns no error, but the search ranks the results much worse. Vectors have 4096 dimensions and come back normalized, so a dot product is the cosine similarity. A request takes up to 64 inputs and 120,000 characters; each input is embedded up to 4,096 tokens, and longer text is cut without an error, so split long documents into chunks first. Your plan’s per-request size limit is for chat and does not apply here, but embedding tokens count toward your hourly token budget and each request uses one of your concurrent slots.
Will the tools I already use work?
Yes, if they speak the OpenAI API. Set the base URL to https://api.reeva.me/v1 and the API key to your Reeva key, then choose a model by name — deepseek-v4.1-flash, qwen-image or qwen3.6-35b-a3b. opencode, Continue, Aider, Cline, the OpenAI SDKs and curl all work this way, without code changes.
Are the tokens really unlimited?
Yes. There is no monthly token cap and no per-token billing — the price is flat. What a plan sets is speed: requests per minute and requests at once. A single request can be up to 32k tokens on Mini and 128k on Starter and Pro, counting the prompt plus max_tokens; a larger request is refused before it runs, with HTTP 400 and the error request_too_large. Scale has no size cap of its own: a request can use DeepSeek v4.1’s full 256k-token context, prompt plus max_tokens. Qwen 3.6 has its own, smaller limit: 32k tokens per request, prompt and answer together, on every plan; a request over it is refused with HTTP 400. Fair use: each plan has an hourly amount of tokens (500k on Mini, 2M on Starter, 6M on Pro and 12M on Scale) and of images (3, 10, 30 and 60), and use above it slows your requests down. Only extreme use sustained over a full day is paused until it eases.
How much does it cost?
Mini is $5 per month, Starter is $29 per month, Pro is $59 per month and Scale is $119 per month — a flat price with no per-token billing. Enterprise pricing is custom.
How do I get an API key?
Choose a plan and pay with PayPal. Your account is created and your first key is issued as soon as the payment clears. In the control panel you can create up to 10 keys, give each one a name and an optional expiry, and revoke any key on its own — a revoked key gets HTTP 401.
How do I pay?
With PayPal, monthly. There is no commitment, and you can cancel at any time in the control panel.
Can I cancel anytime?
Yes. Plans are billed monthly through PayPal and you can cancel in the control panel at any time. Your key keeps working until the period you have paid for ends.
What if I need higher limits?
The Enterprise option offers a dedicated instance, custom rate limits, an SLA and contractual privacy terms. Contact us to discuss your workload.
What happens when I hit a rate limit?
Only four things return HTTP 429: the per-minute limit, a wait of more than two minutes for a free slot, asking for too many images in one request, and the fair-use pause. Over the per-minute limit, retry within a minute, when the window resets. A request over the limit on requests at once waits for one of your own requests to finish, and gets a 429 only if no slot frees up within two minutes. Mini allows 1 request at a time and 5 per minute; Starter allows 2 at a time and 20 per minute; Pro allows 4 at a time and 60 per minute; Scale allows 8 at a time and 120 per minute. One image request can ask for up to 3 images on Mini, 10 on Starter and 16 on Pro and Scale; asking for more is refused at once. The fair-use pause comes only after extreme use sustained over a full day, and lifts as the day rolls on. Going over the hourly amount of tokens or images slows your requests down but does not return a 429.
Do I need a Reeva mail account?
No. The AI API is its own product, on its own subscription, and buying it gets you the API and nothing else. You don’t need an existing Reeva account either: choose a plan, enter an email and a password, and the account is created when the PayPal payment clears. Reeva mail and cloud storage are a separate product you can add later from the same dashboard.
Next step
Choose a plan, create a key, paste it into your tool.
Unlimited tokens from $5 a month, a separate key for every tool, and any key revoked in one click.
Reeva is an independent product and is not affiliated with, endorsed by, or sponsored by any third party mentioned on this page. All product names, logos, and brands are the property of their respective owners and are used here only for identification and comparison.