Pricing & Billing
Byesu aggregates mainstream models through official upstreams — no model downgrade — with pay-as-you-go, transparent multipliers. Text per token, images per image, video per second, cache hits as low as 0.1x, auto-refund on failed media tasks. One token for text, image and video.
Live prices & status on the Model Marketplace
The Model Marketplace shows each model's price, latency and availability live (input / output / cache, by your group when signed in); the console records your actual charge per request. This page is July 2026 static data for search & quick reference — see the Model Marketplace for live prices.
How billing works
| Type | Billing | Notes |
|---|---|---|
| Text / coding | Per token (input + output) | tokens × model multiplier × group multiplier; Claude cache hits as low as 0.1x |
| Image | Per image | charged per image generated |
| Video | Per second | unit price × duration; auto-refund on failure |
💬 Text / coding models
All callable via one OpenAI-compatible interface (Claude also supports the native Anthropic protocol and Claude Code):
| Model | Best for | Billing |
|---|---|---|
| claude-opus-4-8 | Anthropic flagship, strongest reasoning, Effort Control | pay-as-you-go · transparent · cache 0.1x |
| claude-sonnet-5 | General / coding, works with Claude Code | pay-as-you-go · transparent · cache 0.1x |
| claude-haiku-4-5 | Fast & light | pay-as-you-go · transparent · cache 0.1x |
| gpt-5.6 (sol / terra / luna) | OpenAI three-tier family, works with Codex | pay-as-you-go · transparent · cache 0.1x |
| gpt-5.5 | OpenAI previous flagship | pay-as-you-go · transparent · cache 0.1x |
| grok-4.5 | xAI coding / agent model, 500K context | pay-as-you-go · transparent · cache 0.1x |
| gemini-3.1-pro | Google flagship | pay-as-you-go · transparent |
Text prices vary by group. The Model Marketplace shows each model's live price, latency and status (by your group when signed in). Heavy usage plus 0.1x cache hits meaningfully lowers long-context cost.
🖼 Image models (per image)
| Model | About | Price |
|---|---|---|
| gpt-image-2 | OpenAI image | $0.01 / image |
| nano-banana-2 | Google Gemini image | $0.025 / image |
| flux-kontext-max | Flux, instruction edits | $0.025 / image |
| flux-klein-2 | Krea Flux Klein | $0.025 / image |
| firefly-image-5 | Adobe Firefly 5 | $0.025 / image |
| imagine-1.5 | Imagine text-to-image | $0.025 / image |
| seedream-4.5 | ByteDance Seedream | $0.05 / image |
| imagine-1.5pro | Imagine Pro 4K | $0.05 / image |
| firefly-gpt-image-2 | Firefly × GPT, up to 4K | $0.04 / image |
🎬 Video models (per second)
| Model | About | Price |
|---|---|---|
| firefly-video | Adobe Firefly video | $0.015 / sec |
| grok-video | xAI Grok video | $0.015 / sec |
| Veo 3.1 (gemini-veo31) | Google text-to-video, Sora alternative | $0.01875 / sec |
| firefly-ray | Firefly Ray | $0.02 / sec |
| runway-gen4-turbo | Runway Gen-4 Turbo (needs first frame) | $0.025 / sec |
Example: a 6-second Veo 3.1 clip ≈ $0.1125; auto-refund on failure. Image / video tokens need the
media / media-gengroup — see Choosing a Group.
Why the pricing is worth it
- Official upstreams, no downgrade — official channels, not reverse-engineered or mirrored; model specs match the originals.
- Transparent pricing — the console shows actual billed price vs official list price, live; no fuzzy multipliers.
- Cache hits 0.1x — Claude cache hits as low as 0.1x; real savings for long-context heavy usage.
- Refund on failure — image / video tasks auto-refunded on failure.
- One token, all models — text, image and video on one token.
- Local payment — Alipay / WeChat / USDT, invoices available.
FAQ
How does Byesu API billing work?
Text is billed per token (input + output × model & group multiplier); images per image; video per second (unit price × duration). All prices show the actual billed price in the console, compared live against the official list price.
How much do cache hits save?
For Claude models, cache hits are billed as low as 0.1x, matching official cache pricing. For repeated long-context reads, this significantly lowers real cost.
Am I charged if generation fails?
No. Media tasks are auto-refunded on failure.
How do I set up tokens for image and video models?
Image / video models are only available under the media / media-gen group; create a token in that group.
Where do I see the actual price?
The Model Marketplace shows each model's live price, latency and status (by your group when signed in); the console records your actual charge per request. This page is July 2026 static data; the Marketplace is authoritative for live prices.
Get started: Quick Start (3 steps) · Models · Image / Video · Sign up