Pricing

Prepaid credits. No subscription. No markup math.

Buy a credit pack, spend it at your own pace across every model and modality. Every call returns meta.price.usd — the amount you were actually charged, even when it's zero.

Create an account → Per-model rates
Credit packs

Pick a pack. The full sticker amount lands as credit.

Card processing fees are added on top of the pack price rather than taken out of your credit — so a $20 pack puts exactly $20.00 on your balance. The charged amount is what you'll see on your card statement.

starter
$5 credited
  • $5.46 charged → $5.00 credited
  • Credits never expire
  • Every ability on every token
Buy $5 credits
basic
$20 credited
  • $20.91 charged → $20.00 credited
  • Credits never expire
  • Every ability on every token
Buy $20 credits
pro
$50 credited
  • $51.80 charged → $50.00 credited
  • Credits never expire
  • Every ability on every token
Buy $50 credits
power
$100 credited
  • $103.30 charged → $100.00 credited
  • Credits never expire
  • Every ability on every token
Buy $100 credits
How billing works

One rule per modality. All of them visible before you call.

T

Text: published per-model rate

Chat and reasoning calls are billed per 1K input and output tokens at the rate published for that model on the catalog and in GET /v1/models. Rates are synced from provider price lists daily and don't move between calls.

▣

Images: flat per image

$0.06 standard · $0.22 premium, regardless of size or prompt length. Premium applies to Flux 2 Max, GPT Image 1, GPT Image 1.5, Seedream V5 Pro and Seedream V5 Pro Edit. Edits and upscales are priced the same way, per output.

♪

Speech & transcription: metered per call

Text-to-speech is metered on characters, transcription on audio duration — both vary too widely for one honest flat figure, so the exact price comes back in meta.price on every response.

⛨

Prepaid and fail-closed

A token whose account has no remaining balance is refused before any provider is called. A misconfigured account can't generate spend with nowhere to send the bill — and you can't be surprised by a negative balance.

Every response carries meta.price.usd. It is the only cost figure that leaves Fire, and it's the customer-facing charged amount — Fire's own provider cost is never mixed into it.
Per-model rates

53 verified models, every rate published.

Text rates are per 1K tokens. Image models are the flat tiers above. Only onboarded models appear — the same set /v1/models serves.

Chat & reasoning

ModelProviderTierPrice
Claude Sonnet 4.5 default edenai standard $0.0045 in / $0.0225 out per 1K tokens
Venice Uncensored 1.2 venice fast $0.0003 in / $0.00135 out per 1K tokens
Claude Opus 4.5 edenai frontier $0.0225 in / $0.1125 out per 1K tokens
GPT-4o Mini edenai fast $0.00023 in / $0.0009 out per 1K tokens
GPT-5 Mini edenai fast $0.00038 in / $0.003 out per 1K tokens
Gemini 2.5 Pro edenai frontier $0.00188 in / $0.015 out per 1K tokens
DeepSeek Chat edenai fast $0.00041 in / $0.00165 out per 1K tokens
DeepSeek Reasoner edenai fast $0.00083 in / $0.00329 out per 1K tokens
GPT-4o (Azure) edenai standard $0.00375 in / $0.015 out per 1K tokens
GPT-3.5 Turbo edenai nano $0.00075 in / $0.00225 out per 1K tokens
o4 Mini edenai fast $0.00165 in / $0.0066 out per 1K tokens
Gemini 2.5 Flash edenai fast $0.00045 in / $0.00375 out per 1K tokens
Mistral Large (latest) edenai standard $0.00075 in / $0.00225 out per 1K tokens
o3 Mini edenai fast $0.00165 in / $0.0066 out per 1K tokens
GPT-4o edenai standard $0.00375 in / $0.015 out per 1K tokens
Codestral (latest) edenai standard $0.0015 in / $0.0045 out per 1K tokens
GPT-4.1 (2025-04-14) edenai standard $0.003 in / $0.012 out per 1K tokens
Claude Haiku 4.5 edenai fast $0.0012 in / $0.006 out per 1K tokens
Qwen 3.8 27B (Uncensored) venice standard $0.00068 in / $0.0048 out per 1K tokens
Venice Role Play Uncensored venice standard $0.00075 in / $0.003 out per 1K tokens
GLM 4.7 Flash Heretic venice fast $0.00011 in / $0.0006 out per 1K tokens

Image generation

ModelProviderTierPrice
Lustify v7 venice standard $0.06 per image
WAI Illustrious (Anime) venice standard $0.06 per image
Stable Diffusion XL edenai fast $0.06 per image
Flux 2 Pro venice standard $0.06 per image
Flux 2 Max premium venice frontier $0.22 per image
GPT Image 1 premium edenai frontier $0.22 per image
Z-Image Turbo venice fast $0.06 per image
Venice SD 3.5 venice standard $0.06 per image
Lustify SDXL venice fast $0.06 per image
GPT Image 1.5 premium edenai frontier $0.22 per image
Lustify v8 venice standard $0.06 per image
Seedream V4.5 venice standard $0.06 per image
Krea 2 Turbo venice fast $0.06 per image
Qwen Image venice standard $0.06 per image
Chroma venice fast $0.06 per image
Seedream V5 Lite venice fast $0.06 per image
Seedream V5 Pro premium venice frontier $0.22 per image
Qwen Image 3 venice standard $0.06 per image
Qwen Image 3 Pro venice standard $0.06 per image

Image editing

ModelProviderTierPrice
Qwen Edit Uncensored venice standard $0.06 per image
Qwen Image 3 Edit venice standard $0.06 per image
Qwen Image 3 Pro Edit venice standard $0.06 per image
Seedream V4 Edit venice standard $0.06 per image
Seedream V5 Lite Edit venice standard $0.06 per image
Seedream V5 Pro Edit premium venice frontier $0.22 per image

Text-to-speech

ModelProviderTierPrice
Kokoro TTS default venice — Metered per call — price returned in meta.price
ElevenLabs Turbo v2.5 (via Venice) venice — Metered per call — price returned in meta.price
Gemini 3.1 Flash TTS venice — Metered per call — price returned in meta.price
Google Cloud TTS (Neural) edenai — Metered per call — price returned in meta.price
OpenAI TTS edenai — Metered per call — price returned in meta.price

Speech-to-text

ModelProviderTierPrice
Whisper Large v3 Turbo (Groq) default edenai fast Metered per call — price returned in meta.price

Decision models

ModelProviderTierPrice
Jev (TypeSafe) default edenai fast —
Cost-plus partners

Transparent pass-through for infrastructure partners.

Accounts are billed at published rates where one exists, otherwise at provider cost plus a disclosed 50% markup — computed fresh on every call, never estimated.

Partners onboarding at a different markup redeem a tier code — minted by Fire's operator, redeemed once via POST /v1/billing/tier/redeem — which binds the markup to the account. Logging, metering and the fail-closed balance gate are identical to retail.

An open ledger with a single flame rising from its pages beside neatly stacked coins and price tags
FAQ

The questions people ask before they top up.

Can I see what a call will cost before I make it?

Yes. Text rates are published per model on the catalog and in GET /v1/models; images are the flat tiers above. After the call, meta.price.usd tells you what it actually cost.

Do credits expire?

No. Credits stay on your balance until you spend them.

What's the minimum?

The $5 pack. Creating an account and minting a key is free — you only pay when you top up.

Are refunds available?

Credits are prepaid and non-refundable, except where the law requires otherwise. If a call fails on our side you aren't charged for it, so there's nothing to refund.

What happens on a failed call?

A provider error is logged (so you and we can see it happened) but not charged. You only pay for calls that returned a result.

Are there rate limits?

Call reporting is limited to 30/min and ratings to 60/min per token. Inference endpoints run under fair use; if you need sustained high throughput, say so when you sign up.

Ready when you are

Create an account, mint a key, top up when you're ready.

No card required to get your first key.

Get your API key → Read the docs