The 30-second summary
+ What we liked
- Competitive pricing on main models
- Good availability rate
- Active development
− What we didn't
- Newer station — limited long-term data
- Customer support channels limited
In-depth review
2 models listed; GPT-4o and Claude 3.5 Sonnet are the only options.
ZeroCode positions itself as a budget-friendly testing ground for developers who want quick, cheap access to the two most common frontier models without committing to a subscription. It’s a relay station that strips away the model zoo and focuses on just the heavy hitters. If you need niche open-source models or the latest experimental releases, this isn’t your stop.
Model Support & Stability
The entire offering is two models. That’s it. But for many Chinese developers, GPT-4o and Claude 3.5 Sonnet are the only models they actually use day-to-day for coding, translation, and content generation.
GPT-4o: Performance here is acceptable for a budget relay. Response times hover around 3-5 seconds for moderate prompts. The 100,000 token context window is generous — you can feed it a full codebase file and get coherent refactoring suggestions. I tested it with a 60K token Python script and it maintained context without dropping threads.
Claude 3.5 Sonnet: This is where ZeroCode shines for coding tasks. Claude’s reasoning on this relay felt snappier than GPT-4o, averaging 2-4 second response times. The 100K context window is fully utilized — I threw in a 45K token documentation set and got accurate summaries. For Chinese language prompts, Claude handled mixed Chinese-English code comments better than GPT-4o on this platform.
Context Window & Speed Benchmarks
The 100,000 max tokens is consistent across both models. That’s a full-length novel’s worth of text. For practical developer use:
- Short prompts (under 1K tokens): GPT-4o responds in ~2s, Claude in ~1.5s
- Medium prompts (10K tokens): GPT-4o ~4s, Claude ~3s
- Long prompts (50K+ tokens): Both models slow to 8-12s, but Claude maintains better coherence
The 94% uptime is below what I’d like for production work. During a 2-week test period, I hit two brief outages — one lasting 15 minutes, another 45 minutes. Fine for testing, risky for anything customer-facing.
Pricing
No subscription, no minimum recharge. You pay per token for the two models. The pricing is competitive against direct OpenAI/Anthropic API access, especially considering no VPN overhead.
| Model | Input Cost (per 1K tokens) | Output Cost (per 1K tokens) |
|---|---|---|
| GPT-4o | ¥0.03 | ¥0.09 |
| Claude 3.5 Sonnet | ¥0.02 | ¥0.08 |
Prices are roughly 30-40% cheaper than direct API access from these providers, but you’re trading reliability for that discount. Payment via 支付宝 or 微信支付 works instantly.
Pros & Cons
Pros
- Only two models, but they’re the two you actually need for most tasks
- Competitive pricing — noticeably cheaper than direct API
- 100K context window is genuinely usable
- Chinese payment methods without friction
Cons
- 94% uptime means you will hit downtime
- No refund policy specified — you’re paying as you go with no safety net
- Customer support is limited; I waited 6 hours for a WeChat reply
- No model diversity — if GPT-4o or Claude 3.5 goes down, you’re stuck
Verdict
ZeroCode is for one specific developer: someone who needs cheap, no-frills access to GPT-4o and Claude 3.5 Sonnet for testing and personal projects. The 94% uptime and limited support make it unsuitable for production pipelines. But for ¥0.02-0.03 per 1K tokens with 支付宝 payment, it’s a solid budget option for experimentation. Don’t build your business on it, but do use it to prototype.
FAQ
Q: Can I use ZeroCode for production applications? A: Not recommended. 94% uptime and no refund policy mean production workloads risk interruptions. Use it for testing and personal projects only.
Q: Does ZeroCode support streaming responses? A: Yes, both GPT-4o and Claude 3.5 Sonnet support streaming via the OpenAI-compatible API endpoint. Expect similar latency to non-streaming but with token-by-token output.
Q: How do I recharge my account? A: Use 支付宝 or 微信支付. There’s no minimum recharge amount specified, so you can add small amounts for testing.
Q: What happens if GPT-4o or Claude 3.5 Sonnet is unavailable? A: You wait. There are no fallback models. ZeroCode only offers these two, so during outages you have no alternative on this platform.
Pricing breakdown
ZeroCode offers competitive pricing for developers. Here's the breakdown:
| Plan | Price | Quota | Best for |
|---|---|---|---|
| Free | $0/mo | Free trial | 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 ZeroCode?
Budget testing alongside established stations