Talk to it.Test it.Build on it.
Nirvani AI is one API for leading AI models, with streaming, rate limits and live metering built in. Ask it anything here, compare models side by side, then take a free key and build.
Four models. One key.
Pick the tier per request. Same endpoint, same format, same metering.
| Model | Best for | Context | Try it |
|---|---|---|---|
Nirvani Flashnirvani-flash | Fast, capable default for chat, support and agents. | 1M tokens | Playground |
Nirvani Flash Litenirvani-flash-lite | Lowest cost per token for classification, routing and summaries. | 1M tokens | Playground |
Nirvani Pronirvani-pro | Deepest reasoning and writing quality for analysis and complex agents. | 200K tokens | With a key |
Nirvani Opennirvani-open | Open-weights model for transparency and steady pricing. | 128K tokens | Playground |
Change one line. Keep your code.
Nirvani AI speaks the standard chat-completions format. Point your base URL at Nirvani, add your key, and your existing SDK, framework or no-code tool keeps working.
curl https://api.nirvani.ai/v1/chat/completions \
-H "Authorization: Bearer $NIRVANI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "nirvani-flash",
"stream": true,
"messages": [{"role": "user", "content": "Draft a follow-up text for a missed call"}]
}'const res = await fetch("https://api.nirvani.ai/v1" + "/chat/completions", {
method: "POST",
headers: {
Authorization: `Bearer ${process.env.NIRVANI_API_KEY}`,
"Content-Type": "application/json"
},
body: JSON.stringify({
model: "nirvani-flash",
messages: [{ role: "user", content: "Summarize today's leads" }]
})
});
const data = await res.json();
console.log(data.choices[0].message.content);import os, requests
r = requests.post(
"https://api.nirvani.ai/v1" + "/chat/completions",
headers={"Authorization": f"Bearer {os.environ['NIRVANI_API_KEY']}"},
json={
"model": "nirvani-pro",
"messages": [{"role": "user", "content": "Score this lead from 1 to 10"}],
},
timeout=60,
)
print(r.json()["choices"][0]["message"]["content"])Production controls, included.
The parts every team ends up building around a model API are already here.
- Streaming by default
- Tokens arrive as they are generated, in the standard event-stream format, so replies appear as they are written.
- Per-key rate limits
- Every key carries its own requests-per-minute ceiling. A runaway script never takes down the rest of your stack.
- Live metering
- Requests, tokens, latency and errors per key and per model, updated as calls land. Real numbers, viewable in the console.
- Monthly token quotas
- Spend stays predictable. Each key has a monthly token budget and a clear 429 when it is reached.
- Keys stored as hashes
- Your key is shown once, by email. Nirvani keeps only a SHA-256 hash, so a database copy exposes nothing usable.
- Global edge network
- Requests are handled at the edge closest to your users. There are no servers to run or scale.
From first question to full AI team.
Start in the chat. Stay as long as it keeps paying for itself.
Talk to it
Ask the assistant above anything, or have it play your receptionist for a minute.
Test it
Send one prompt to several models in the playground and compare speed and quality.
Build on it
Take a free test key, change one URL, and ship your first feature.
Grow with it
Go live with production limits, or let Nirvani run your AI Receptionist, AI Sales Rep and AI Follow-Up Specialist for you.
Questions, answered.
What is Nirvani AI?
Nirvani AI is one API for leading AI models, built by Nirvani Company. It adds the controls a business needs in production: streaming, per-key rate limits, monthly token quotas and live usage metering. Talk to it here, compare models in the playground, and build on it with a free test key.
Does it work with my existing code?
Yes. The API follows the standard chat-completions format. Most existing SDKs and tools work after you set the base URL to https://api.nirvani.ai/v1 and use your Nirvani key.
How much does it cost?
Test keys are free and include 20 requests per minute and 100,000 tokens per month. Live keys for production volume are set up per business: reply to your key email and the team will set one up.
Which models can I use?
Four tiers: Nirvani Flash for fast everyday work, Nirvani Flash Lite for the lowest cost at high volume, Nirvani Pro for the deepest reasoning and writing, and Nirvani Open for open-weights workloads. GET /v1/models returns the live list.
How are API keys protected?
Keys are delivered once by email and stored only as SHA-256 hashes. Each key has its own rate limit and monthly quota, and its usage is visible in the console.
What happens when I reach a limit?
The API answers HTTP 429 with a clear message: rate_limit_error when you exceed requests per minute, or quota_exceeded when the monthly token quota is used. Live keys raise both limits.
Your free key is one email away.
20 requests per minute and 100,000 tokens a month, on every model tier. It lands in your inbox in seconds.