Transparent. Wholesale. No surprises.
One predictable invoice. Per-token rates published openly. Three plans — Pro, Scale, and Enterprise — each with a transparent monthly platform fee and bulk discounts on every model.
Pro
For teams shipping AI to production with real volume.
- 15–25% bulk discount on every model
- All 31 frontier models, one API
- Smart routing & automatic fallbacks
- Team workspace · 5 seats included
- Per-key budgets & rate limits
- Email support · 1-business-day reply
Scale
For high-traffic AI products and platform teams.
- 25–45% bulk discount, volume-based
- Quality-aware routing & semantic cache
- Audit logs · prompt-level observability
- 25 seats · multi-project workspaces
- PII redaction & guardrails
- Priority support · 4-hour reply SLA
Enterprise
For organizations with custom compliance and volume.
- 40–60% bulk discount, contract-locked
- Dedicated capacity · private gateway / BYOC
- EU-only / US-only data residency
- SOC 2, DPA, MSA, custom SLAs
- Named account manager · 24/7 on-call
- White-glove migration & integration
See what you save, model by model.
Prices shown as input / output, in USD per 1M tokens. Updated weekly.
| Model | Direct provider | Blue Box | You save |
|---|---|---|---|
| GPT-5 | $10.00 / $40.00 | $8.40 / $33.60 | −16% |
| Claude Opus 4 | $15.00 / $75.00 | $10.50 / $52.50 | −30% |
| Claude Sonnet 4 | $3.00 / $15.00 | $2.40 / $12.00 | −20% |
| Gemini 2.5 Pro | $1.25 / $10.00 | $1.05 / $8.40 | −16% |
| Llama 4 Maverick | $0.50 / $1.80 | $0.30 / $1.10 | −40% |
| Mistral Large | $2.00 / $6.00 | $1.50 / $4.50 | −25% |
| DeepSeek V3 | $0.27 / $1.10 | $0.18 / $0.72 | −34% |
Embedding, fine-tuning, and image models follow the same wholesale model. Full price list →
Pricing questions, answered.
Still curious? Talk to a human →
Is this legal? How do you offer wholesale rates?
Yes. We operate under signed reseller agreements with select providers, route open-weight models to the cheapest verified hosts, and aggregate enterprise tier-purchasing. Every credit is sourced from a compliant, verifiable channel — never gray-market accounts.
Do I need to change my code?
Almost never. We are OpenAI-compatible, so any existing client (LangChain, LlamaIndex, AI SDK, etc.) works by changing the base URL and API key. Provider-specific features are exposed via our extended schema.
Can I bring my own provider keys (BYOK)?
Yes. On the Scale and Enterprise plans you can attach your own provider keys and still benefit from routing, observability, fallbacks, and budget controls. Useful if you have existing committed-use discounts.
How is the platform fee structured?
Each plan includes a flat monthly platform fee (Pro $499 · Scale $4,999 · Enterprise custom) plus the wholesale per-token rate of every model you use. The platform fee includes seats, observability, audit logs, support and routing — no surprise add-ons.
Is there a minimum commitment?
Pro and Scale are month-to-month, cancel any time. Enterprise contracts are typically annual with quarterly true-ups — terms are negotiated directly.
What about data privacy?
We never train on your data. Prompt logging is opt-in and granular. EU customers can pin all traffic and storage to EU regions. SOC 2 Type II is in audit; DPAs are signed on request.
Refunds and credits?
Unused balance can be refunded any time within 90 days of purchase. We absorb provider outages — if a fallback succeeds, you only pay for one call.
Run the math. Then run the migration.
Send us your last 30 days of model usage and we'll send back a per-model savings report within 24 hours.