APIKEY.FUN API Review 2026: Claude & GPT for China
APIKEY.FUN review: 40+ models (Claude Code, GPT, Gemini, DeepSeek), China-direct, ¥1=$1 pricing, and how it compares to OpenRouter and FreeModel.
TL;DR: APIKEY.FUN is a China-direct API reseller exposing 40+ models (Claude Code, GPT-4o, Gemini, DeepSeek, Qwen, Kimi) via an OpenAI-compatible endpoint. ¥1=$1 transparent pricing, WeChat support, no proxy required. Best for China-based developers running Claude Code or Codex. Less ideal for international multi-model research (use OpenRouter) or mission-critical workloads (run a backup provider).
Introduction: An API Reseller Built for China
APIKEY.FUN is a Chinese API reseller that wraps Claude, GPT, Gemini, DeepSeek, Qwen, Kimi, Doubao, and Zhipu GLM behind a single OpenAI-compatible endpoint. The pitch is simple: one account, one bill, no proxy, no developer friction.
What makes APIKEY.FUN interesting in 2026 is its focus on Claude Code and OpenAI Codex tool-calling — two of the most popular agentic coding workflows in 2026 — plus a pricing formula that's actually transparent ($1 = ¥7, with a clear group multiplier). For developers in mainland China who need Anthropic or OpenAI models without setting up a proxy, this is a legitimate alternative to OpenRouter and FreeModel.
This review covers APIKEY.FUN's model lineup, pricing structure, China access reality, and how it stacks up against the other major API aggregators.
APIKEY.FUN Pricing Breakdown
APIKEY.FUN uses a "group" pricing model. Each model belongs to a group with a multiplier. The formula is simple:
Final price = Official price × Group multiplier ÷ 7
The ÷7 comes from the $1 = ¥7 exchange assumption — meaning if you recharge in RMB, the resulting USD cost matches what you'd pay at the official provider.
| Group | Models | Typical Multiplier | Example |
|---|---|---|---|
| Group 1 | DeepSeek V3, Qwen, GLM, Kimi, Doubao | 1.0–1.5× | Cheapest tier |
| Group 2 | Claude Sonnet/Haiku, GPT-4o-mini | 1.5–2.0× | Mid tier |
| Group 3 | Claude Opus, GPT-4o, o1, o3 | 2.0–3.0× | Premium tier |
| Group 4 | Gemini 2.5 Pro, specialty models | 1.5–2.5× | Mid-premium |
How Much Can You Get for ¥100 (≈$14)?
| Use Case | Models | Approximate Volume |
|---|---|---|
| DeepSeek V3 chat | Group 1 (1.0×) | ~14M input tokens |
| Qwen 2.5 chat | Group 1 (1.0×) | ~14M input tokens |
| Claude Sonnet coding | Group 2 (1.5×) | ~1.5M input tokens |
| GPT-4o coding | Group 3 (2.0×) | ~700K input tokens |
| Claude Opus agent | Group 3 (2.5×) | ~140K input tokens |
Free Tier
APIKEY.FUN offers a small signup credit for new users — enough to test the API across multiple models. There's no monthly free quota like OpenRouter's free tier; the signup credit is a one-time allocation that lets you verify connectivity and pricing before committing to a recharge.
Key Advantages of APIKEY.FUN
- China-direct access: No proxy, no VPN. The endpoint resolves from mainland China without any extra setup. Critical for Claude Code and Codex workflows that need stable connectivity.
- OpenAI-compatible endpoint: Drop-in replacement for the OpenAI SDK. Most existing code works with just an API base URL change.
- Claude Code + Codex support: First-class tool-calling support for the two most popular agentic coding CLIs in 2026. This is the main reason many Chinese developers choose APIKEY.FUN.
- Transparent pricing formula: $1 = ¥7 with a public group multiplier. You can calculate the final cost before recharging.
- Chinese interface + Chinese support: WeChat-based customer service (Laoye1999eth), Chinese-language dashboard, Chinese billing.
- 40+ models: All major Chinese and international models under one account — no juggling multiple API keys.
Limitations to Consider
- Group multiplier opacity: The exact multiplier per model is not always published in advance. You may need to recharge a small amount to confirm the final cost on specific models.
- Less known than OpenRouter: OpenRouter is the global standard for API aggregation. APIKEY.FUN's English documentation is thinner, and the model catalog updates lag behind OpenRouter by days to weeks.
- No public pricing page: There's no canonical "pricing" page you can bookmark. Pricing lives in the dashboard, which is Chinese-first.
- Recharge friction: Top-up happens via WeChat transfer (Laoye1999eth) or Alipay, not credit card. International users will find this awkward.
- Single point of failure: If APIKEY.FUN has an outage, every model on your account is affected. Diversifying across providers (e.g., APIKEY.FUN + FreeModel + direct OpenAI) is the safer pattern.
APIKEY.FUN vs OpenRouter vs FreeModel
| Factor | APIKEY.FUN | OpenRouter | FreeModel |
|---|---|---|---|
| China access | ✅ Direct | ❌ Proxy required | ✅ Direct |
| Claude Code | ✅ Native | ⚠️ Limited | ✅ Native |
| Codex support | ✅ Native | ⚠️ Limited | ✅ Native |
| Pricing transparency | Medium (group formula) | High (per-model public) | High (per-model public) |
| Payment | WeChat/Alipay | Credit card, crypto | Credit card, Alipay |
| Model count | 40+ | 200+ | 30+ |
| Language | Chinese-first | English-first | Bilingual |
| Reliability | Single region, smaller team | Multi-region, larger team | Multi-region |
| Best for | China + Claude Code | International, max model choice | China + international, balanced |
Use Case Recommendations
| Use Case | Recommended | Why |
|---|---|---|
| China-based Claude Code user | APIKEY.FUN | Native tool-calling + no proxy |
| International multi-model research | OpenRouter | 200+ models, transparent pricing |
| China + international mixed | FreeModel | Direct access, bilingual, DeepSeek partnership |
| Anthropic-only workflow | APIKEY.FUN or direct Anthropic | Choose based on CN access needs |
| OpenAI Codex agent | APIKEY.FUN or direct OpenAI | Both work; APIKEY.FUN simpler for CN |
How to Get Started
- Sign up: Visit apikey.fun and create an account (Chinese phone or email works).
- Recharge: Top up via WeChat (Laoye1999eth) or Alipay. Minimum recharge is typically ¥10–¥50.
- Generate API key: Dashboard → API keys → Create new key.
- Configure Claude Code / Codex: Replace the OpenAI base URL with the APIKEY.FUN endpoint, paste your key, and start coding.
- Test spend: Run a small batch of test calls to confirm the group multiplier for the models you actually use.
FAQ
Q: Is APIKEY.FUN the same as OpenRouter?
No — OpenRouter is a global aggregator with 200+ models and English-first documentation. APIKEY.FUN is a China-first reseller with 40+ models, Chinese dashboard, and WeChat-based support. Both expose OpenAI-compatible endpoints.
Q: Can I use Claude Code with APIKEY.FUN?
Yes — Claude Code is one of the main use cases. Set the OpenAI base URL to APIKEY.FUN's endpoint, use any Claude model, and tool-calling works natively.
Q: Is APIKEY.FUN cheaper than direct OpenAI/Anthropic?
It depends on the model and group multiplier. For Group 1 models (DeepSeek, Qwen), APIKEY.FUN is at parity with direct pricing. For Group 3 (Claude Opus, GPT-4o), the multiplier makes APIKEY.FUN 1.5–3× more expensive than direct — but it saves the proxy and friction cost.
Q: Does APIKEY.FUN work outside China?
Yes, the API endpoint is accessible globally. International users will find the Chinese-first dashboard awkward but the API itself works the same.
Q: Is APIKEY.FUN reliable?
It's a smaller team with a single-region deployment. For mission-critical workloads, run APIKEY.FUN alongside a backup provider (e.g., FreeModel or direct OpenAI) and failover if the primary goes down.
Q: What's the minimum recharge?
Typically ¥10–¥50 depending on the payment method. There's no monthly subscription — it's pure pay-as-you-go.
Conclusion
APIKEY.FUN is a pragmatic choice for China-based developers who need Claude Code, Codex, or international models without proxy hassle. The transparent ¥1=$1 formula and WeChat support are real advantages. The tradeoffs are group multiplier opacity and a smaller team than OpenRouter.
If you're building agentic coding workflows from China, APIKEY.FUN is worth a ¥50 test recharge. For pure research or non-China development, OpenRouter remains the better default. If you want a balance between the two — direct access, bilingual, broader model coverage — check out FreeModel, which is the China-direct partner of DeepSeek's official pricing.
Comparison Table
| Provider | China Direct | Models | Best For |
|---|---|---|---|
| APIKEY.FUN | ✅ | 40+ | Claude Code, China devs |
| OpenRouter | ❌ | 200+ | International, max choice |
| FreeModel | ✅ | 30+ | CN + international, balanced |
| Direct Anthropic | ❌ | Claude only | Enterprise, no proxy ok |
| Direct OpenAI | ❌ | OpenAI only | Enterprise, no proxy ok |