One plan.
Priced like it's 2026.
Every seat is one person with unlimited interactive inference on models up to 70B. That's the whole pricing page — but here's the proof.
Plus applicable tax. Cancel anytime. Switch to annual and pay $11/year instead.
A seat is one person. Signing up a team? See Enterprise for volume seats.
- Unlimited interactive inference
- Every model, up to 70B
- OpenAI-compatible API + web playground
- Streaming, function calling, JSON mode
- Commercial use included
- Transparent Fair Use — no token billing
Compare against a metered bill
Drag to your real usage. See what per-token pricing would cost the same team.
Flat seats vs. the token meter
The same workload, two billing philosophies. One of them makes you ration your own team.
| What you get | Metered API | Denizen Blu |
|---|---|---|
| Monthly bill | Varies with every token | Seats × $1.99 — fixed |
| Surprise overages | Common at scale | Impossible by design |
| Models up to 70B | Extra per-token tier | All included |
| Rate-limit anxiety | Throttle to save money | Build freely |
| Streaming & function calling | Usually metered too | Included |
| Commercial use | Depends on tier | Included |
| Forecasting next quarter | Spreadsheet guesswork | Multiply by $1.99 |
| Web playground | Often paid add-on | Included |
Enterprise, when you need more than a seat
Everything in the Unlimited Seat, plus the controls and capacity larger teams require. Volume seat pricing available.
- SSO / SAML, SCIM provisioning, audit logs
- Dedicated capacity and higher concurrency
- Priority support with an SLA and onboarding
Answered plainly
A seat is one person. You pay $1.99 per month (or $11 per year), and that person gets unlimited interactive inference across every model. Add or remove seats anytime; changes are prorated.
Correct. There is no per-token, per-request, or per-model charge. Your bill is seats × $1.99/month (or × $11/year). The Fair Use Benchmark governs scheduling priority for extreme outliers, never billing.
Yes — 7 days on the full Unlimited Seat, no credit card required to start. You only enter billing details if you decide to keep going.
Yes. Annual billing is the same flat model and saves you roughly two months versus paying monthly.
Nothing changes on your invoice. If a single seat runs far above the 50M-token benchmark, that seat may receive slightly reduced scheduling priority during peak demand. Sessions are never terminated and there are no hard caps.
Yes. Commercial and production use is included. For high-concurrency production traffic or dedicated capacity, enterprise plans add the headroom and SLAs you'll want.
Stop counting tokens
Give your team a seat and let them build like the meter doesn't exist — because it doesn't.