No token billing. Ever.

Unmetered AI inference.
Where intelligence belongs.

Run open models up to 70B without token meters, GPU rentals, or unpredictable API bills. One seat. One price. Unlimited interactive use.

For people, not bots. Transparent Fair Use applies. No credit-card gymnastics.

$1.99Per seat / month
70BMax model size
0Tokens counted
Interactive use
Live across the Denizen network 1,248,502 tok/s
12,408Active seats
2,914M tokens today
27Models live
$0Overage billed, ever
Simulated preview data — the live product renders real network telemetry here.

Teams building without the meter

◈ NorthwindPalettely◎ Corvid LabsRunlevel✦ AdastraBytemark
The end of token billing

Intelligence shouldn't be rationed

Metering made sense for electricity. It makes no sense for thinking. Denizen Blu is built so a person working all day, every day, never has to think about tokens again.

No token billing

Stop counting inputs and outputs. Build, prototype, and ship without a mental tax on every request.

Any model up to 70B

The full open catalog — Llama, DeepSeek, Qwen, Mistral, and more. You choose.

OpenAI-compatible API

Point your existing code at a new base URL and keep moving. No rewrites, no new SDK to learn.

Built for teams

Centralized billing, seat management, shared API keys, and usage visibility from one clean dashboard.

Streaming, always

Token-by-token streaming responses and a built-in web playground for fast iteration and demos.

Transparent fair use

A published benchmark, no hard monthly caps, and no mid-session cutoffs. You always know where you stand.

See the difference

What the meter really costs you

Drag to match your usage. Watch what a per-token bill does — and what one flat seat does instead.

Your usage

A rough model based on a typical 70B-class API rate of ~$0.80 per million output tokens.

LightTypicalHeavy
Metered API (per token)$200
Denizen Blu (flat seat)$100
$100/mo
saved with unmetered pricing
Illustrative estimate for interactive use, not a quote. Fair Use applies.
One price, everyone belongs

Built for the way you actually work

Ship without watching the meter

Prototype loops, retries, big context windows, agent chains — the things that make token bills scary — are just Tuesday here.

  • Swap in with a single base-URL change
  • Streaming, function calling, JSON mode
  • A web playground for fast iteration
Read the docs
chat.py
# Same SDK. New base URL. No token math.
from openai import OpenAI

client = OpenAI(
    base_url="https://bluroute.denizenblu.com/v1",
    api_key="db-•••••••••••",
)

resp = client.chat.completions.create(
    model="llama-3.3-70b-instruct",
    messages=[{"role": "user",
               "content": "Ship it."}],
    stream=True,
)  # bill unchanged: $1.99/seat

Budgets you can predict to the dollar

Seats × $1.99/month (or × $11/year). That's the whole invoice. No end-of-month surprises, no reconciling usage dashboards, no throttling your own engineers to save money.

  • Centralized billing and one invoice
  • Add or remove seats in seconds
  • Per-seat keys and usage visibility
Set up your team
invoice.txt
# This month's forecast
Seats            12
Rate             $1.99 / seat / mo
──────────────────────────
Token overages   $0
Surprise bills   $0
──────────────────────────
Total            $240.00
# same every month. by design.

Scale intelligence across the company

When AI is unmetered, adoption stops being a line item you police and starts being something you encourage. Give every team member a seat and let usage grow.

  • SSO, admin controls, and audit visibility
  • Dedicated capacity and higher concurrency
  • Enterprise support and onboarding
Talk to sales
adoption.log
# Adoption when nobody watches the meter
Q1   ████████░░░░░░░░   220 seats
Q2   ████████████░░░░   610 seats
Q3   ███████████████░   1,400 seats
Q4   ████████████████   2,900 seats
# cost per seat: still $1.99.
Your choice of models

Any open model you choose

You choose what thinks for you. Every model is served behind the same API, at the same flat seat price.

One API, every model

Switch between any model in the catalog by changing a single string — no new endpoints, no new billing tier, ever.

The open catalog

Choose your own inference — Llama, DeepSeek, Qwen, Mistral, and more, up to 70B. Don't see one? Request it.

Request any model

The catalog is community-driven. Missing a model? Request it and vote — popular open models go live within days.

Browse the full catalog →
API quickstart

Drop-in OpenAI compatibility

Keep your OpenAI SDK. Point the base URL at Denizen Blu, drop in your key, and choose any model — the whole migration is three lines.

Python
Node.js
cURL
from openai import OpenAI

client = OpenAI(
    base_url="https://bluroute.denizenblu.com/v1",
    api_key="db-xxx",
)

resp = client.chat.completions.create(
    model="llama-3.3-70b-instruct",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(resp.choices[0].message.content)
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://bluroute.denizenblu.com/v1",
  apiKey: "db-xxx",
});

const resp = await client.chat.completions.create({
  model: "llama-3.3-70b-instruct",
  messages: [{ role: "user", content: "Hello!" }],
});
console.log(resp.choices[0].message.content);
curl https://bluroute.denizenblu.com/v1/chat/completions \
  -H "Authorization: Bearer db-xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "llama-3.3-70b-instruct",
    "messages": [{"role":"user","content":"Hello!"}]
  }'
Pricing

One plan. No asterisks.

One seat, one person, unlimited interactive inference.

Unlimited SeatMOST POPULAR
$1.99/ seat / month

or $11/seat/year, plus applicable tax

Get started
  • Unlimited interactive inference
  • Models up to 70B
  • OpenAI-compatible API + web playground
  • Streaming responses & commercial use
  • Transparent Fair Use — no token billing

Need SSO, dedicated capacity, or enterprise support? Talk to sales →

Transparent by design

Unlimited means unlimited — for one person

Denizen Blu is designed so a human working all day, every day, never has to think about tokens. Fair Use exists to protect that experience, not to claw it back.

No hard monthly caps and no mid-session cutoffs — sessions are never terminated.
Usage far above the benchmark may see reduced scheduling priority at peak — never a bill.
For people, not bots. One seat is one person, not a shared firehose.
Read the full policy →
50Mtokens / mo
The published Fair Use Benchmark — a reference point, not a wall.
Questions

The obvious ones, answered

Unlimited interactive use by one person. If you're a human working through a playground or your own app all day, you'll never hit a wall or a bill. The Fair Use Benchmark (50M output tokens/month) is a published reference so heavy users know roughly where scheduling priority may ease during peak demand — sessions are never cut off and there are no hard caps.

Your invoice is seats × $1.99/month (or × $11/year). It doesn't move based on how many tokens you send or receive. You never reconcile a usage dashboard, buy credits, or get a surprise overage. The price you see is the price you pay.

If you already use an OpenAI-compatible SDK, it's a base-URL and API-key change. Streaming, function calling, and JSON mode all work the way you expect. Most teams are running in minutes.

Whichever you want, up to 70B. Your seat includes every open-weight model in the catalog — Llama, DeepSeek, Qwen, Mistral, and more. Missing one? Request it and vote — popular models go live within days. Everything is included at the same flat seat price.

Yes. Commercial use is included in every seat. Build products, ship features, and run it in production — the seat price covers it.

Yes. For SSO, admin controls, dedicated capacity, higher concurrency, and enterprise support, talk to sales and we'll tailor a plan to your organization.

AI without the meter

Intelligence shouldn't be rationed, and it shouldn't be metered. Give your team a seat and let them think freely.