BYOK · HK Edge · Zero Markup

AIMACTOK

Route overseas AI calls to Chinese models through Hong Kong edge. Your keys. 89ms latency. MCP-native.

89ms
Edge Latency
37%
Latency Cut
0%
Markup
MCP
Native
// TL;DR

AimActok at a glance

17
Providers

DeepSeek, MiniMax, Qwen, Doubao, GLM, Moonshot, plus OpenAI / Anthropic / Google. One OpenAI-compatible endpoint.

89ms
p50 latency

Hong Kong edge cuts the US→China hop. p99 under 250ms on most providers.

0%
Token markup

Pay providers directly. Software subscription is the only fee.

MCP
Native protocol

Works with Claude Code, Cursor, Continue, Cline out of the box.

1k/mo
Free tier

No credit card. Pro $7.9 · Max $29 · Team $109.

// Architecture

How It Works

Three layers of intelligence between your app and Chinese AI models

01

BRING YOUR
OWN KEYS

Connect your existing API keys from any Chinese model provider. DeepSeek, Qwen, GLM, Moonshot, MiniMax — bring what you already have. Zero key storage on our side.

02

HK EDGE
ROUTING

Your calls hit our Hong Kong edge nodes — the fastest path from global traffic into China's AI infrastructure. Sub-100ms guaranteed.

03

MCP-NATIVE
GATEWAY

Native MCP protocol support. Drop our config into your MCP client and all your Chinese models appear as tools. No SDK required.

// Performance

The Proof

Measured. Transparent. Verifiable.

Direct — US West
250-350ms
US West → China Mainland direct route
Aimactok — HK Edge
89ms
US West → HK Edge → China Mainland

MCP Configuration

// mcp.json — drop into any MCP client
{
  "mcpServers": {
    "aimactok": {
      "url": "https://edge.aimactok.io/mcp",
      "env": {
        "AIMACTOK_KEY": "sk-your-key-here",
        "PROVIDER": "deepseek"
      }
    }
  }
}
Claude Desktop VS Code Copilot Cursor Windsurf Continue.dev Zed Neovim Emacs
DeepSeek
V3 / R1
Qwen
2.5 / Max
GLM
4 / 4V
Moonshot
Kimi
ByteDance
Doubao
Zhipu
AutoGLM
// AI Tools Directory

All Tools. One Gateway.

Beyond chat. Test, benchmark, and route any AI tool through Aimactok edge.

New
Gateway
BYOK Router
Edge-routed inference at P50 89ms across 14 providers. Zero markup, your keys.
Benchmark
Latency Dashboard
Real-time P50/P95 latency, last 24h, all regions. Verified by users.
MCP
MCP Server
Drop-in MCP config. Works with Claude, Cursor, Windsurf, Zed, Neovim.
Security
Key Vault
Bring-your-own-key model. We never store your provider credentials.
Analytics
Usage Analytics
Per-model cost, request count, error rate. Export to CSV.
Edge
Global Edge
HK · SG · US · EU. Sub-100ms guaranteed for top-tier plans.
// Field Reports

What Developers Ship Faster

1,847 indie devs and small teams shipping AI products through Aimactok edge.

"We cut Claude Code's response time in half by routing Chinese models through Aimactok. Setup was 3 lines of mcp.json. No SDK, no boilerplate."

M
Marcus T.
Indie Dev · Singapore

"Our Cursor tab completions went from 'laggy' to 'instant' after switching to HK edge. 240ms down to 90ms is a 2.6x speedup users actually feel."

K
Kenji O.
Full-stack · Tokyo
// Access Tiers

Pick Your Edge

All plans include HK edge routing, MCP-native access, and zero markup on tokens. You pay providers directly.

STARTER
$0/ MONTH
1,000 requests / mo
Try the edge · No card
  • HK edge routing
  • Low latency dashboard
  • Usage stats & CSV
  • 1 API key
DEPLOY FREE
MAX
$29/ MONTH
50,000 requests / mo
Same as Pro, scaled up
  • HK edge routing
  • Low latency dashboard
  • Usage stats & CSV
  • Wallet overage $0.001/call
  • 1 API key
DEPLOY MAX
TEAM
$109/ MONTH
200,000 requests / mo shared
For teams & studios
  • Everything in Pro
  • 20 API keys (shared vault)
  • Dedicated edge nodes
  • SLA 99.9% uptime
  • SSO + audit logs
  • Priority Discord channel
DEPLOY TEAM

PAY WITH PAYPAL · CARD · USDC · CANCEL ANYTIME · NO LOCK-IN

// Latest Articles

Fresh from the blog

Benchmarks, BYOK deep-dives, and AI tool reviews — updated weekly.

BENCHMARK · 2026-06-27

BYOK Latency Benchmark: 17 Providers Tested

p50/p95/p99 across 4 model families. The fastest gateway is not the one you think.

Read benchmark →
PRICING · 2026-06-27

Cheapest AI Image Generation API in 2026

Replicate, MiniMax, Midjourney v7, DALL-E 3, Stable Diffusion — real costs per image.

Read comparison →
PRICING · 2026-06-07

Subscription vs PAYG vs BYOK: Honest Pricing

BYOK wins for anyone using 2+ models. The 1–3% margin math, with real numbers.

Read breakdown →
REVIEWS · 2026-06-07

Best AI Tools in 2026: Honest Reviews

22 AI tools I actually pay for in 2026. What they cost, what they do, which ones survive 30 days.

Read reviews →
All articles →

READY TO CUT LATENCY?

Deploy in 5 minutes. Your keys. Zero markup. Pure speed.

DEPLOY NOW