An OpenAI-compatible AI API gateway for global developers. DeepSeek, GLM, Kimi, Qwen, MiniMax — one endpoint, pay-as-you-go, self-serve signup with free trial quota.
Upstream disclosed · dual-provider auto-failover · prompts not used for training · no regional limits
Get an API Key — Sign up free Read the QuickstartPoint any OpenAI SDK at our endpoint and pick a model:
export OPENAI_API_KEY=sk-xxxx
export OPENAI_BASE_URL=https://keheai.com/v1
# Python
from openai import OpenAI
client = OpenAI(base_url="https://keheai.com/v1", api_key="sk-xxxx")
resp = client.chat.completions.create(
model="deepseek-chat",
messages=[{"role": "user", "content": "Hello!"}],
)
print(resp.choices[0].message.content)
# curl
curl https://keheai.com/v1/chat/completions \
-H "Authorization: Bearer sk-xxxx" \
-H "Content-Type: application/json" \
-d '{"model":"deepseek-chat","messages":[{"role":"user","content":"Hello!"}]}'
Flagship lightweight chat — great cost/performance balance.
Reasoning model for complex tasks, math, and analysis.
High-efficiency general model (DeepSeek V3.2).
Strong agentic flagship from Zhipu AI.
Code-specialized model for programming workloads.
Balanced all-rounder.
Long-context and MoE options.
Embeddings, images, speech, rerankers. Check /v1/models.
| Model | Input | Output |
|---|---|---|
| deepseek-chat (V4-Flash) | $0.90 | $2.70 |
| deepseek-chat — Night Pack (02:00-08:00 UTC+8) | $0.45 | $1.35 |
| deepseek-v3 (V3.2) | $0.60 | $0.90 |
| deepseek-reasoner (V4-Pro) | $1.90 | $3.80 |
| Kimi-K2.7-Code | $1.05 | $4.30 |
| GLM-5.2 | $1.30 | $4.50 |
Monthly packs available for steady workloads. Email us for volume pricing.
Tell us what you are building — we usually reply within hours (UTC+8 business hours).