Built for OpenClacky

Top-Tier Model Access, Optimized for Your Workflow

高速系列（DeepSeek V4 Pro / Flash）深度优化。稳定路由、智能缓存、零加价。

Pay-as-you-go · No subscription · Auto top-up supported

Why OpenClacky AI Keys

One Key, every top-tier model

DeepSeek series — all in one place. Best reasoning × best value, without juggling multiple accounts, keys, and bills.

One Key, every model

Unified access to major models — DeepSeek. Switch by alias, top up once, use everywhere.

Reasoning + value + vision, all-in-one

DeepSeek V4 Pro leads deep reasoning, DeepSeek V4 Flash nails cost-efficiency.

Stable, low latency, official pricing

Globally reachable proxy at official rates (only 5% service fee). Near-100% cache hit cuts long-conversation cost by ~50%.

Beyond chat · Fully multimodal

One Key for text, images, video and voice

Same Key, same balance, same Base URL. Code, draw, shoot video and narrate from one top-up — no more signing up on every vendor's platform.

Chat & reasoning

DeepSeek and other leading models for code, reasoning and long documents.

DeepSeek V4 Pro · V4 Flash
POST /chat/completions

Image generation & editing

Text-to-image, image editing and multi-image compositing — posters to product shots in one prompt.

Text-to-image · Editing · Compositing
POST /images/generations

Video generation

Text-to-video and first-frame-to-video with native audio, cinematic 720p / 1080p.

Text-to-video · First-frame-to-video
POST /videos/generations

Speech & transcription

Multi-voice TTS narration and speech-to-text — give your videos and agents a voice.

TTS · Speech-to-text
POST /audio/speech
Image showcase

One prompt, one piece

Every image below was generated with a single AI Keys call, no retouching. About $0.035 per image.

Image API docs →
Rainy neon Tokyo alley, girl in a translucent raincoat glancing back, puddle reflections, 35mm film look Cinematic photo
“Rainy neon Tokyo alley, girl in a translucent raincoat glancing back, puddle reflections, 35mm film look”
Swiss-style poster titled ONE KEY, violet-to-orange gradient geometry Poster design
“Swiss-style poster titled ONE KEY, violet-to-orange gradient geometry”
Close-up portrait with glowing bioluminescent patterns, navy and magenta Hyper-real portrait
“Close-up portrait with glowing bioluminescent patterns, navy and magenta”
Colossal whale carrying a glowing ancient city through a sea of clouds, golden hour, god rays Fantasy art
“Colossal whale carrying a glowing ancient city through a sea of clouds, golden hour, god rays”
Futuristic floating crystal perfume bottle, prismatic rainbow caustics, minimal dark backdrop Product shot
“Futuristic floating crystal perfume bottle, prismatic rainbow caustics, minimal dark backdrop”
Miniature late-night programmer studio, sleeping orange cat on the desk, warm light Isometric 3D
“Miniature late-night programmer studio, sleeping orange cat on the desk, warm light”
Video showcase

An 8-second cinematic shot, one API call

A prompt in, a 720p clip with audio out — no editing software. This one took a single AI Keys call and cost under $0.50.

Prompt

“Slow drone shot through a futuristic neon city canyon, flying cars leaving light trails, light rain and haze, holographic billboards, cinematic camera movement” 

curl https://api.openclacky.com/videos/generations \
  -H "Authorization: Bearer $AI_KEY" \
  -d '{"model": "...",
       "prompt": "...", "duration_seconds": 8}'
8s · 720p · with audio · ≈ $0.47 Video API docs →

Three flagship lineups in one place

Cost-efficient

DeepSeek Series

Peak/off-peak billing: off-peak at half price, weekends off-peak all day. Combined with 99% cache hit rate, long-conversation cost drops another 50%.

DeepSeek V4 Pro · V4.1 Flash
Coding · Multimodal

GLM Series

From Zhipu AI. The flagship GLM-5.3 targets complex coding and long-horizon tasks; GLM-5.3-Flash is the line's first natively multimodal model, beating the previous flagship at roughly one-tenth the price. Direct China access for both.

GLM-5.3 · GLM-5.3-Flash
Long Context

Kimi Series

From Moonshot AI, with direct China access. The flagship Kimi K3 packs a 2.8T-parameter MoE with native vision and 1M context — built for long-horizon coding and deep research.

Kimi K3
Price cut

DeepSeek flash series price cut — V4.1 Flash is much cheaper

From Sep 10, V4.1 Flash bills at the new rates automatically; V4 Pro stays available with unchanged pricing.

Rates below are for the V4.1 Flash series: peak 09:00-12:00 & 14:00-18:00 (Beijing) · off-peak auto half price · weekends off-peak all day.

Input / 1M tokens
Peak $0.30
Off-peak $0.15
Output / 1M tokens
Peak $1.20
Off-peak $0.60
Cache hit / 1M tokens
Peak $0.006
Off-peak $0.003
Create AI Key →

Why Not Just Use Your Own API Key?

Unreliable Connectivity

Direct API connections often suffer high latency, timeouts, and rate limits that break your agent mid-task.

Unpredictable Billing

Token-based pricing from multiple providers is hard to track. Unexpected bills kill momentum.

Compatibility Risk

Different providers use varying API standards, which can cause compatibility issues when switching or mixing them — quietly breaking your Agent's reliability.

How It Works

1

Top Up Your Balance

Add $20 to start. Pay-as-you-go, no subscription.

2

Generate an AI Key

One key for all models. Paste it directly into OpenClacky settings.

3

Use the Best Models

Enjoy optimized, cached access to top models with transparent per-request pricing.

Why AI Keys?

Stable, Optimized Routing

Direct connections to top-tier cloud infrastructure, tuned specifically for OpenClacky workloads. No middleman hops, no fragile tunnels. Sub-second response times instead of minute-long waits.

Response latency
Minutes
Sub-second

Zero Markup

You pay exactly what the model providers charge. No reseller margin. No hidden fees. Just 5% for payment processing. No monthly subscription.

Our markup
0%

Industry-Leading Cache Hit Rate

While other gateways achieve ~90% cache hit rates, AI Keys reaches 99%. Every cache hit saves 90% of the token cost. The result: up to 50% lower overall spend without changing a single line of your workflow.

Cache hit rate
Other gateways 90%
AI Keys 99%
Cost reduction
50%↓

Frequently Asked Questions

What is AI Keys?
AI Keys is a managed model access service built for OpenClacky. One key gives you access to DeepSeek (V4-Pro / V4.1-Flash) — with stable routing and smart caching.
How is it different from buying API keys directly?
Direct API keys require separate accounts, different configurations, and individual billing for each provider. AI Keys gives you one unified key, one balance, and one pricing table.
Which models are supported and how do I switch?
Currently the DeepSeek series — full list and pricing at /docs/ai-key-models. Inside OpenClacky just switch by model alias; no need to reconfigure your key.
Are there any promotions?
New users get free credit on sign-up to try things out. Some models run limited-time promotions (e.g. the DeepSeek flash series price cut) — announced in the console and docs.
What is cache hit rate? How does it cut cost by another 50%?
DeepSeek and other models support prompt caching: identical prefixes are billed at as low as 10% of normal input rates on hit. In agent workflows with repeated system prompts the hit rate can reach 99%, cutting overall input cost by ~50%.
What does it cost?
You pay the official model rates with no markup. We add 5% to cover payment processing. With near-100% cache hits the total cost of long conversations drops another ~50%. No monthly fee, no hidden charges.
Is my data safe?
Requests are forwarded to providers with no data retention on our side. We do not use your data for training.
Can I set a spending limit?
Yes. You can configure a monthly spending cap from your Creator Hub at any time.

Ready to Start?

Top up $20 and get your AI Key in under a minute.

Top Up Now →

Already have an account? Sign in