GPT-4o input costs roughly $2.50 per million tokens via Aiberm, and Claude 3.5 Sonnet runs about $6 per million input tokens — roughly 2x the official Anthropic API price. Aiberm is a multi-provider relay station targeting Chinese developers who want GPT-4o and Claude 3.5 Sonnet without a VPN. It supports Alipay and WeChat Pay, making it convenient for local users. But the pricing story is more complicated than the headline rates.
Pricing Breakdown
Aiberm’s pricing is not transparently published per-million-token. Based on the platform data, here is what you get:
| Model | Input Cost (per million tokens) | Output Cost (per million tokens) | Notes |
|---|---|---|---|
| GPT-4o | ~$2.50 | ~$10.00 | Estimated from usage |
| Claude 3.5 Sonnet | ~$6.00 | ~$18.00 | ~2x official price |
There is no free trial. The free_trial field is False. You must recharge via Alipay or WeChat Pay before using the API. |
|||
| Key takeaways: |
- Claude pricing is significantly higher than direct API — 2x markup is steep.
- No free trial means you pay upfront.
- Payment via Chinese wallets is a plus for local users.
Hidden Costs and Minimum Top-Ups
Two things generic pricing articles miss: batch discount thresholds and CNY minimum top-up amounts. Aiberm does not specify a minimum recharge amount in CNY. This is a red flag. Without a floor, you could theoretically deposit ¥1, but the platform may impose a practical minimum. No batch discount thresholds are mentioned either — you pay the same per-token regardless of volume. There are no hidden per-request overhead fees listed in the data. But the 2x markup on Claude suggests you are paying a convenience premium. Key takeaways:
- No batch discounts available.
- Minimum top-up is unspecified — potential friction for small users.
Feature Matrix vs OpenRouter
| Dimension | Aiberm | OpenRouter |
|---|---|---|
| Models | GPT-4o, Claude 3.5 Sonnet only | 200+ models |
| Max tokens | 100,000 | 128,000 |
| Uptime | 99.3% | 99.5%+ |
| Latency | 1308ms | ~800ms |
| Payment | Alipay, WeChat Pay | Credit card, crypto |
| Free trial | No | $1 credit |
| OpenRouter wins on model variety, uptime, and latency. Aiberm wins on Chinese payment methods. | ||
| Key takeaways: |
- Aiberm is slower and less reliable than OpenRouter.
- If you need Chinese payment, Aiberm is the better choice.
Pros & Cons
Pros:
- Broad model support beyond the two listed: Kimi, MiniMax, DeepSeek, GLM, Grok
- Operator has multiple proven projects — not a fly-by-night operation
- Stable operation on most channels Cons:
- Claude pricing at ~2x official — not the cheapest option
- Limited differentiation from competitors like OpenRouter or Helpaio
- No standout feature for Claude users specifically
- No free trial, no batch discounts, no minimum top-up transparency
Verdict
Aiberm is a solid option if you need Chinese payment methods and want access to GPT-4o and Claude 3.5 Sonnet without a VPN. But the 2x markup on Claude stings, and the lack of a free trial or minimum top-up clarity adds friction. For developers who prioritize cost, direct API access or OpenRouter is cheaper. For developers who prioritize convenience and local payment, Aiberm works. Bottom line: Good for Chinese developers who need quick access and local payment, but not the cheapest or fastest option.
FAQ
Q: Does Aiberm offer a free trial?
A: No. The platform data shows free_trial: False. You must recharge via Alipay or WeChat Pay before using the API.
Q: What is the minimum recharge amount in CNY?
A: Not specified in the platform data. You may need to contact support or attempt a deposit to find out.
Q: Are there batch discounts for high-volume usage?
A: No batch discount thresholds are mentioned in the platform data. Pricing appears to be flat per-token.
Q: How does Aiberm’s latency compare to OpenRouter?
A: Aiberm averages 1308ms latency, while OpenRouter is around 800ms. Aiberm is slower.
Q: Can I use Aiberm without a VPN in China?
A: Yes. That is the primary use case. Aiberm is designed for Chinese developers who need access to GPT-4o and Claude 3.5 Sonnet without a VPN.
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.
