Pricing & Credits
Brody AI charges in credits, not dollars per message. Here is exactly how credits are spent, what every plan includes, and what happens when a call fails.
What is a credit?
Credits are a token-based unit of AI usage. Each request reserves 1,500 credits (~5,000 tokens), and the final charge is calculated from the tokens actually used:
credits_charged = max(250, ceil((tokens_in + 2 × tokens_out) × 0.3))
- Roughly 0.3 credits per token — output tokens count double.
- Every call costs at least 250 credits, even very short ones.
- Unused reservations are refunded automatically when the call completes — you are only charged for what was consumed.
There is no fixed USD-to-credit conversion: provider cost is tracked in dollars per call for transparency, and the free model pool costs $0 at the provider — your credits cover gateway infrastructure.
Plans & allotments
| Plan | Price | Monthly credits | Calls | Notes |
|---|---|---|---|---|
| Trial | $07 days | 50,000 credits | 100 calls | Granted at signup. No credit card required. |
| GO | $9/month | 50,000 credits | 1,000 calls | Standard models, priority processing. |
| Pro | $19/month | 100,000 credits | 2,500 calls | Frontier models (Claude, Gemini, Nemotron). |
| Studio | $29/month | 250,000 credits | 5,000 calls | For full-time solo developers. |
| Team | $79/month | 500,000 credits | 20,000 calls | Shared across a small team. |
| Enterprise | $99/month | Unlimited | Unlimited | Custom deployment and support. |
Source of truth: backend plan catalog (backend/src/billing/paymentService.ts). A $10 credit pack (100,000 credits) is also available as a one-off top-up — pay-as-you-go credits never expire.
Model cost per 1M tokens
USD figures come straight from the router's provider cost matrix. These are what the upstream providers charge — used for the per-call cost ledger. Your credit balance is burned per token at the flat rate above, regardless of which model answers.
| Model | Input / 1M | Output / 1M | Tier |
|---|---|---|---|
| Free pool (:free models, openrouter/free) | $0.00 | $0.00 | Free |
| cohere/north-mini-code:free | $0.00 | $0.00 | Free |
| google/gemini-3.5-flash | $0.00115 | $0.009 | Premium |
| google/gemini-3.6-flash | $0.0015 | $0.0075 | Premium |
| google/gemini-2.0-flash-001 | $0.10 | $0.40 | Premium |
| google/gemini-pro | $0.50 | $1.50 | Premium |
| openai/gpt-4o-mini | $0.15 | $0.60 | Premium |
| openai/gpt-4o | $5.00 | $15.00 | Premium |
| anthropic/claude-3-5-haiku | $0.80 | $4.00 | Premium |
| anthropic/claude-3.5-sonnet | $3.00 | $15.00 | Premium |
| deepseek/deepseek-chat | $0.14 | $0.28 | Premium |
| deepseek/deepseek-r1 | $0.55 | $2.19 | Premium |
Failed calls are free
If a call fails at the system level (upstream outage, timeout, canceled mid-stream), the reserved credits are refunded in full automatically — refunds are idempotent per request, so a retried or duplicated request can never double-charge you.
You never pay for a response you did not receive.
Safety net
Under a total upstream outage the gateway streams a local high-demand notice instead of failing silently. That notice is marked as a failed call — it is free and the reservation is refunded. A higher-tier paid safety net is in flight.
Tip: bring your own key
With BYOK you use your own provider key (OpenAI, Anthropic Claude, Google Gemini, DeepSeek, Ollama, OpenRouter, Groq, or a custom endpoint) and pay the provider directly at their published rates. See the models page for the full provider list.
No card for trial
Unused credits never expire (packs)
Auto-refunds on failure
Credit top-ups available