Pricing · overview

primerouter charges per token, real-consumption billing — no monthly fee, no plan, no minimum.

Core principles

1. Per-token billing: actual tokens consumed × model price = cost
2. No inflation: usage field is byte-identical to upstream
3. No bundle prepay: top up = quota; consume what you use; remainder lasts forever (unless account closed)
4. Pre-charge + actual settlement: long requests pre-charge by estimated max, settle by real usage on completion, refund difference

Detailed rules in Billing rules.

Three price-affecting dimensions

Final charge = Model baseline price  ×  Model multiplier (system-wide)  ×  Group multiplier (your group)

• Model baseline price: see Model price table
• Model multiplier: default 1.0; some models have markup or discount
• Group multiplier: default 1.0; VIP / partner have special multipliers, see Groups

Example:

Item	Value
Baseline (qwen-max input)	$2.00 / 1M tokens
Model multiplier	1.0
My group (vip-customer)	0.8
Actual	$1.60 / 1M tokens

Multimodal pricing

Each modality has its own billing dimensions:

Modality	Billed by
Text chat	prompt + completion tokens
Image generation	count × per-image price
Video generation	duration × per-duration price (or frames)
Music generation	duration × price
TTS	character count × per-char price
STT	audio seconds × per-second price
Embeddings	input tokens
Rerank	candidate count × per-call price

Specific prices in Model price table.

Crypto top-up conversion

Crypto deposits convert at credit-time rate:

• USDT / USDC / etc. stablecoins: fixed 1 USD
• Native tokens (ETH / BNB / APT): real-time rate (oracle-provided by site config)

See Crypto Payment.

Further reading

• Model price table — every model + live price
• Groups — group multipliers, plans
• Billing rules — pre-charge / settle / refund / edge cases