The 30-second summary
+ What we liked
- Code-focused relay with good Claude support
- Competitive pricing
- Active updates
− What we didn't
- Newer station with limited history
- Focused on coding use cases only
In-depth review
Overview
aicodemirror positions itself as a niche relay for developers who primarily use Claude Code and GPT-4o for coding tasks. It’s a newer entrant in the AI token relay space, targeting Chinese developers who want direct access to these models without VPN configuration. The platform is deliberately focused — it doesn’t try to be everything to everyone, and that’s both its strength and limitation.
Pricing & Models
aicodemirror offers a straightforward model lineup: GPT-4o and Claude 3.5 Sonnet. There’s no free tier, but the pricing is competitive compared to direct API access from OpenAI or Anthropic.
| Model | Pricing | Max Tokens | Notes |
|---|---|---|---|
| GPT-4o | Pay-per-use (no fixed monthly) | 100,000 | Good for chat and code generation |
| Claude 3.5 Sonnet | Pay-per-use (no fixed monthly) | 100,000 | Strong for complex code reasoning |
The platform does not advertise a specific monthly subscription fee — it appears to be usage-based. The 100K token context window is standard for these models, and sufficient for most coding tasks.
China Access & API Compatibility
For developers behind the Great Firewall, aicodemirror works without VPN. You can access both models via a standard OpenAI-compatible API endpoint, which means existing tools like OpenCat, ChatBox, or custom scripts can connect with minimal configuration changes.
The relay handles the routing to OpenAI and Anthropic servers, so you don’t need to manage API keys from the original providers. Just register, top up credits, and start coding.
Uptime & Reliability
Uptime is listed at 94.0%, which is below average for relay stations in this space (most competitors target 99%+). This is a notable concern — if you’re relying on this for daily development work, you may experience occasional downtime. The platform is actively updated, which suggests the team is working on improvements, but the track record is still short.
Safety Rating: 4/5
The safety rating is solid but not perfect. Data is not stored or logged in a way that compromises user privacy, but as a newer station, the security auditing is less mature than established players like API2D or AiHubMix.
Pros & Cons
Pros
- Code-focused relay: Optimized for Claude Code and GPT-4o workflows, not general chatbot use.
- Competitive pricing: Usage-based billing often cheaper than direct API calls, especially for heavy users.
- Active updates: The platform receives regular improvements and bug fixes.
Cons
- Newer station with limited history: No long-term reliability data. You’re betting on a young service.
- Focused on coding use cases only: If you need multimodal models (DALL-E, vision), or other providers (Gemini, DeepSeek), this isn’t your relay.
- Sub-94% uptime: Not ideal for production or time-sensitive work.
Verdict
aicodemirror is a solid choice if you primarily use Claude 3.5 Sonnet and GPT-4o for coding and want a no-VPN relay in China. The pricing is fair, the setup is straightforward, and the focus on code means fewer distractions.
However, the 94% uptime and short history are real risks. If you’re building a production pipeline that needs 99.9% reliability, look elsewhere. For personal projects, experimentation, or as a secondary relay, it’s worth trying.
Who should use it: Developers who live in Claude Code and GPT-4o, want a simple relay without model bloat, and can tolerate occasional downtime.
Who should skip: Anyone needing vision models, Gemini access, or mission-critical uptime.
FAQ
Q: Does aicodemirror work without a VPN in China?
A: Yes, the relay is accessible directly from mainland China. You don’t need any additional tools — just register and use the API endpoint.
Q: Can I use aicodemirror with existing tools like OpenCat or Cursor?
A: Yes, the API is OpenAI-compatible. You can plug it into any tool that supports custom API endpoints. Just set the base URL to aicodemirror’s endpoint and use the provided API key.
Q: What happens if the uptime drops below 94%? Do I get compensated?
A: The platform does not advertise any uptime SLA or compensation policy. Given the 94% uptime figure, you should not rely on this for production-critical workflows without a fallback.
Q: Does aicodemirror support streaming responses?
A: Yes, both GPT-4o and Claude 3.5 Sonnet support streaming via the relay. This works out of the box with compatible clients.
Q: How do I top up credits? What payment methods are accepted?
A: The platform supports common payment methods for Chinese users, including Alipay and WeChat Pay. You top up credits and usage is deducted per request.
Pricing breakdown
aicodemirror offers competitive pricing for developers. Here's the breakdown:
| Plan | Price | Quota | Best for |
|---|---|---|---|
| Free | $0/mo | Limited | Kicking the tires |
| Standard RECOMMENDED | Pay-as-you-go/mo | Unlimited usage | Solo devs · small teams |
| Enterprise | Custom | SLA · dedicated support | Teams & agencies |
Supported models
2 models across major vendors.
Frequently asked questions
Can I access this platform from China without a VPN?
Most relay stations are accessible from Chinese ISPs. Check our review for specific routing details.
What payment methods are accepted?
Payment options vary by platform. Some accept Alipay/WeChat Pay, others are USD/crypto only.
How does this compare to using OpenAI directly?
Relay stations add routing latency but provide access from restricted regions, unified billing, and multi-model fallback.
Is my API key safe?
Keys are encrypted at rest. Most platforms support per-project scoping and IP allow-lists.
Should you use aicodemirror?
Claude Code users wanting variety