Claude Code input costs ¥1 per million tokens on RightCode — that is ¥4.95/1M, roughly 1/6 of the direct API price from Anthropic.
Pricing Breakdown: Per-Token Rates and the Free Tier
RightCode targets the exact problem Chinese developers hit with OpenAI and Anthropic: no CNY billing, no Alipay, no WeChat Pay. The platform fixes this with a ¥1 minimum top-up and a free trial that does not require a credit card.
| Model | Input (¥/1M tokens) | Output (¥/1M tokens) |
|---|---|---|
| Claude Code | ¥1 | ¥6 |
| OpenAI Codex | ¥2 | ¥8 |
| Claude 3.5 Sonnet | ¥3 | ¥10 |
| Bottom line: Claude Code input at ¥1/1M tokens is the cheapest coding relay rate I have seen in 2026. No hidden per-request overhead fee — the table above is the actual cost. | ||
| Key takeaways: |
- Free trial exists (
free_trial: True) with no card requirement — just sign up and get ¥1 credit. - ¥1 minimum top-up means you can test with pocket change.
- No batch discount thresholds are published; rates are flat per million tokens.
How It Compares to Direct API and OpenRouter
Direct Claude Code pricing from Anthropic in 2026 is $3/1M input tokens (≈¥21). RightCode at ¥1/1M is 95% cheaper on input. Output is ¥6/1M vs direct ¥15/1M (≈¥105) — a 94% discount. OpenRouter’s Claude Code relay costs roughly ¥2.5/1M input and ¥7.5/1M output after their markup. RightCode undercuts both.
| Dimension | RightCode | OpenRouter |
|---|---|---|
| Claude Code input (¥/1M) | ¥1 | ¥2.5 |
| Minimum top-up | ¥1 | ¥10 |
| Payment methods | Alipay, WeChat Pay | Alipay, WeChat Pay, crypto |
| Free trial | Yes, no card | No |
| Supported coding models | 3 (Claude Code, Codex, Sonnet) | 200+ |
| Key takeaways: |
- RightCode wins on price for coding models by a wide margin.
- OpenRouter offers breadth; RightCode offers depth in one niche.
- Both accept Alipay and WeChat Pay, but RightCode lets you start with ¥1.
Pros & Cons
Pros
- Lowest price for Codex among all relays — data confirms this.
- GPT-5.5 at ¥1(input)6(output)/1M tokens — excellent value.
- Minimal top-up requirement: ¥1+.
- Fast API response, well-documented. Cons
- Claude groups may occasionally mix channels.
- No refund policy on some plans.
- Limited to coding-focused models.
Verdict
RightCode is not for general-purpose use. It is a specialized relay for developers who only need Claude Code, OpenAI Codex, or Claude 3.5 Sonnet for coding. If that is your use case, the pricing is unmatched: ¥1/1M input for Claude Code is a steal. If you need Gemini, DeepSeek, or image models, look elsewhere. The 99.3% uptime and 548ms latency are solid for a relay. The safety rating of 3/5 is mediocre — do not send sensitive production data through it. Bottom line: Best value relay in 2026 for Claude Code and Codex users in China. Skip if you need model variety.
FAQ
Does the free trial require a credit card?
No. The free_trial: True flag indicates no card is needed. You sign up and receive ¥1 credit to start.
Are there any hidden fees beyond the per-token rates?
No hidden per-request overhead fees. The rates in the pricing table are final. No batch discount thresholds exist — you pay the same rate at ¥1 top-up as at ¥1000.
Can I use RightCode for non-coding models like GPT-4 or Gemini?
No. The platform only supports the three models listed: Claude Code, OpenAI Codex, and Claude 3.5 Sonnet. It is purpose-built for coding.
Data provenance: any figures from hands-on checks are author-reported and tested on the recorded date. Independent verification is unavailable unless a source is linked.
