New Sign up free, 10 calls on us. Up to $1, no card needed.

Pricing · updated 2026-05-29

Pay-as-you-go.
Zero surcharge, no subscription.

Top up your balance via Stripe. Zero platform fee on any request. Zero BYOK surcharge. Stripe processing fee on us. Promo top-ups stack additional discounts. No monthly minimum, no per-seat fees, no plan tiers.

Platform key

Zero platform fee

  • We hold keys to Anthropic / OpenAI / Google / others
  • Top up your balance via Stripe; pay only for tokens consumed
  • Zero surcharge on every request
  • Cache reads at ~10% of normal input price
Get your API key →

BYOK

0% BYOK surcharge

  • Add your own provider keys to the workspace vault
  • Keep using your existing provider account
  • Synthorai surcharge waived (0%)
  • Per-model whitelist, fallback toggle, audit log
Add my keys →

$50 launch top-up

$50 → 10% off 30 days

  • Fund $50, get extra 10% off platform-key requests
  • 30-day discount window, balance never expires
  • Stacks with normal usage
  • Refundable within 7 days
Claim promo →

High volume? Additional discounts apply beyond list price: talk to us about volume.

Why it's cheaper than an aggregator

A $100 top-up, same models, same provider list prices across chat, image and audio. The gap is fees.

Synthorai0% fees
$100.00
of usage
OpenRouter~8% fees
$92.59
after 5.5% platform + 2.5% payment
+8%
more usage, before caching
Your numbers vary by model & cache rate. Try the cost calculator →

Sample prices

Per million tokens, USD. Exactly what each request draws from your balance, zero surcharge added. A spread across the catalogue, pulled live at build time.

Provider Model Input ($/Mtok) Output ($/Mtok)
Alibaba qwen3-vl-flash $0.05 $0.4
Amazon nova-2-sonic $0.06 $0.24
ByteDance Dola-Seed-2.0-mini $0.1 $0.4
Anthropic claude-fable-5 $10.00 $50.00
OpenAI gpt-5.4-pro $30.00 $180.00
Google google-tts-chirp3-hd $30.00 $0

Compare all 113 model prices, live → Filter by provider, sort by price, always in sync with the gateway.

How billing actually works

Top up a wallet via Stripe; requests draw it down at the provider's per-token price.

Under the hood every request follows the same path: we pre-deduct an estimate, forward the call, then settle against the usage the provider reports. If a request fails mid-stream, an inflight ledger releases the hold automatically, and the difference is refunded rather than swallowed. A crash on our side must never corrupt your balance; that rule is enforced in code and audited in public: gateway cache-billing audit.

Your balance never expires, and the unused part is refundable within 7 days.

What “0% BYOK” actually means

Bring your own provider keys and requests bypass our wallet entirely: you pay Anthropic, OpenAI, or Google directly at list price, and Synthorai adds nothing on top. That is 0%, permanently, not a promo.

Keys live in a per-workspace encrypted vault. You can whitelist them per model, and a “never fall back to the platform key” flag guarantees traffic only uses your key. If it runs out, the request fails loudly instead of silently billing your wallet.

BYOK stays free because those teams still use caching, failover, audit logs, and workspaces, and their platform-key overflow traffic tends to come through us anyway.

How caching changes your bill

Cache reads cost roughly 10% of input price; cache writes about 125% (Anthropic's schedule; we pass provider pricing through unchanged). Cross-provider translation is documented in our prompt caching guide.

The part that matters in practice: caching only pays off if repeat requests land on the upstream that holds the cache. Synthorai pins cache-enabled traffic to one upstream key per workspace, so a 60%-cacheable workload cuts ~50% off input spend. Estimate yours with the LLM cost calculator.

FAQ

Is there a subscription or monthly minimum?

No. Pay-as-you-go. Top up your balance once, requests draw from it as you use them. Unused balance carries over indefinitely.

Are there any platform fees or surcharges?

None. Zero platform fee on any request. Zero BYOK surcharge. Stripe processing fee on us. Your balance is debited exactly by what each request consumes.

How does prompt caching pricing work?

Cache reads cost ~10% of input price (Anthropic schedule). Cache writes (first call within TTL) cost ~125% of input. We pass these through directly. Cross-provider caching translation is documented in Prompt Caching Across LLM Providers.

What about refunds?

Unused balance refundable within 7 days, no questions asked. Stripe processes the refund directly. We keep your balance correct even if requests die mid-flight. For billing you can verify, see our gateway cache-billing audit.

How does Synthorai compare to OpenRouter?

Same model list prices, but without the extra fees: no payment-processing surcharge, and BYOK is free. See the full OpenRouter alternative comparison.

Is there a free tier?

New accounts get 10 trial calls with up to $1 in free credit, no card required. After the trial, Synthorai is pay-as-you-go with zero platform fee, and BYOK carries a 0% surcharge.

Do prices differ from going to the provider directly?

No. Per-token prices mirror each provider's list price, and the catalog is synced from provider pricing at build time. You pay the same rate; you just stop paying extra fees.

What happens when my balance runs out mid-request?

The in-flight request completes and settles, since we do not cut streams mid-response. New requests are rejected until you top up; the final settlement can leave the balance slightly negative.

Start with whatever fits

Free signup, fund $1 if you want to try, top-up promo when you're ready.