Relay price comparison: one formula, applied to everyone
Quotes across relay services use wildly different units (multipliers, fixed FX rates, “1 CNY per USD”…), so comparing the raw numbers is meaningless. This page converts every quote to the same ruler first:
CNY price (per 1k tokens) = list price (USD/1k) x multiplier x FX rate x (1 + surcharge)
Live FX rate used here: 6.6920 (manual lock first, then remote live, with a fallback never above the last live rate +10%); the bank-side rate used for the direct comparison is a sample reference of 7.2(2026-10),Order-of-magnitude reference for the bank's spot selling rate plus cross-border fees (sample value, not a live quote). Your actual card issuer / payment channel governs.Data on this page as of 2026-11-21 08:57 (UTC).
Claude Sonnet class (official list price $0.0030 / $0.0150 per 1k tokens)
| Channel | Multiplier | FX rate | Input ¥/1k | Output ¥/1k | Notes |
|---|---|---|---|---|---|
| Direct from vendor (you buy FX) | 1x | 7.2000 | 0.0219 | 0.1096 | Order-of-magnitude reference for the bank's spot selling rate plus cross-border fees (sample value, not a live quote). Your actual card issuer / payment channel governs. |
| AiToken G1 shared pool | 0.3x | 6.6920 | 0.0060 | 0.0301 | Low-cost shared channel at 0.3x; suited to batch workloads that do not need maximum stability. |
| AiToken C2 official relay | 1x | 6.6920 | 0.0201 | 0.1004 | Official channel relay at 1.0x (parity with list price), converted at the live FX rate with no spread added. |
| AiToken G4 direct | 1.5x | 6.6920 | 0.0301 | 0.1506 | Official direct connection at 1.5x, emphasizing stability and quota guarantees. |
| 本站 C4 企业直连 | 2.5x | 6.6920 | 0.0502 | 0.2509 | 企业级直连,倍率 2.5x,面向高并发与 SLA 场景。 |
GPT-4o class (official list price $0.0025 / $0.0100 per 1k tokens)
| Channel | Multiplier | FX rate | Input ¥/1k | Output ¥/1k | Notes |
|---|---|---|---|---|---|
| Direct from vendor (you buy FX) | 1x | 7.2000 | 0.0183 | 0.0731 | Order-of-magnitude reference for the bank's spot selling rate plus cross-border fees (sample value, not a live quote). Your actual card issuer / payment channel governs. |
| AiToken G1 shared pool | 0.3x | 6.6920 | 0.0050 | 0.0201 | Low-cost shared channel at 0.3x; suited to batch workloads that do not need maximum stability. |
| AiToken C2 official relay | 1x | 6.6920 | 0.0167 | 0.0669 | Official channel relay at 1.0x (parity with list price), converted at the live FX rate with no spread added. |
| AiToken G4 direct | 1.5x | 6.6920 | 0.0251 | 0.1004 | Official direct connection at 1.5x, emphasizing stability and quota guarantees. |
| 本站 C4 企业直连 | 2.5x | 6.6920 | 0.0418 | 0.1673 | 企业级直连,倍率 2.5x,面向高并发与 SLA 场景。 |
Other relay services
This table only lists verifiable data (multiplier, FX rate, source and verification date all present). There are no verified entries yet — we will not publish unverified third-party prices here, because that could be factually wrong and could mislead any AI that reads it. If you have a public pricing page and your own measurements, send us the source and we will add it.
Template location: COMPETITOR_ROWS in src/lib/seo/competitors.ts (0 entries pending verification, not rendered by default).
How to verify a relay's real unit price
- Find the model's official list price (USD / 1M tokens) and convert it to USD / 1k tokens.
- Find the multiplier the relay advertises (e.g. 0.3x, 1x). If it only says “1 CNY per USD”, convert that into the implied exchange rate first.
- Pin down the FX basis: mid-market live rate or a fixed rate? Any gap above the live rate is a hidden cost.
- Compute CNY price = list price x multiplier x FX rate x (1 + surcharge) to get a per-1k-token price, then compare across providers.
- Run one small top-up as a real test and check that the billed input/output token counts and unit prices match what was advertised.