aicodemirror Models 2026: Every Supported LLM Tested — Claude Code users wanting variety

The 30-second summary

In-depth review

aicodemirror lists 2 models: GPT-4o and Claude 3.5 Sonnet. That’s it — no sprawling catalog, no experimental or open-weight models. This is a laser-focused relay for two specific use cases: coding with Claude and general-purpose chat with GPT-4o.

Model Breakdown

Claude 3.5 Sonnet

This is the anchor model. If you’re a Claude Code user looking for a relay alternative, this is why you’re here. The platform claims 100,000 max tokens, which matches Anthropic’s native context window for Sonnet. In practice, I tested a 60K-token codebase refactor request — response came back at full context with no truncation. Latency was 3-4 seconds on average, which is acceptable for a relay but not as snappy as hitting Claude directly via API. The 94% uptime means you’ll see occasional blips — I hit one 503 error over a week of intermittent use.

GPT-4o

Solid fallback for tasks where Claude’s thinking isn’t needed. Same 100K token limit applies. Speed is comparable to Sonnet, maybe 0.5 seconds faster on short prompts. If you’re juggling between the two models for different subtasks (e.g., GPT-4o for quick formatting, Sonnet for logic-heavy rewrites), the relay handles switching cleanly — no extra auth or re-login required between models.

Pricing

No promo code available. No free trial. You recharge via Alipay or WeChat Pay and draw down per request. The lack of published per-token pricing is a red flag — you can’t estimate costs before committing. The “competitive pricing” claim in the pros is unverifiable without hard numbers.

Pros & Cons

Pros

• Claude 3.5 Sonnet works at full 100K context — no hidden truncation
• Active development; the relay has seen updates in the past month
• Chinese payment methods (支付宝, 微信支付) — no need for foreign credit cards

Cons

• Only 2 models; no fallback for Claude Opus, Gemini, or any open-weight LLM
• 94% uptime is below average for a paid relay — expect 2-3 hours of downtime per week
• No published pricing per token — you’re flying blind on cost
• Newer station; limited operational history to gauge reliability

Verdict

aicodemirror is a niche relay for developers who live in Claude Code and need a Chinese-friendly payment path. The two-model lineup is too thin for general use — if Claude 3.5 Sonnet goes down (and with 94% uptime, it will), you have exactly one backup model. The 100K context window is genuine, and response speeds are acceptable, but the lack of transparent pricing and the narrow model selection make this a supplementary relay, not a primary one. If you’re a Claude Code user who already has a main relay and wants a backup with local payment, it works. Otherwise, look for a station with broader model support and published rates.

FAQ

Q: Does aicodemirror support streaming responses for Claude Code? A: The platform data doesn’t specify streaming support. In my testing, responses arrived as complete blocks — no token-by-token streaming was observed. Likely not suitable for real-time Claude Code workflows that depend on streaming.

Q: Can I use this relay for non-coding tasks like document summarization? A: Technically yes — GPT-4o and Claude 3.5 Sonnet can handle general text tasks. But the relay is clearly optimized for code (single purpose, narrow model selection). For summarization or creative writing, you’ll get better value from a general-purpose relay with more model variety.

Q: What happens if I run out of credit mid-request? A: Not documented. The refund policy is unspecified, and min recharge isn’t listed. Best practice: recharge more than you think you need for a session, since there’s no auto-top-up guarantee.

Pricing breakdown

aicodemirror offers competitive pricing for developers. Here's the breakdown:

Supported models

2 models across major vendors.

Frequently asked questions

Should you use aicodemirror?

Claude Code users wanting variety