ZKA Model — AI Inference API

Documentation: zeroknowledge0x.github.io/zka-model/

ZKA Model is the Model Layer of the ZKA Labs ecosystem — a high-performance AI inference platform that provides fast, reliable access to large language models through a RESTful API. It is designed for production workloads, offering low-latency completions, streaming responses, and flexible authentication.

---

Available Models

Model	ID	Description	Best For
ZK-V3Z	zk-v3z	Flagship reasoning model — highest accuracy and largest context window	Complex reasoning, code generation, research tasks
ZK-V2B	zk-v2b	Balanced model — strong quality with faster response times	General-purpose chat, content creation, summarization
ZK-V1B	zk-v1b	Lite model — ultra-fast inference with minimal latency	High-throughput pipelines, classification, simple Q&A

Tip: Use zk-v1b for latency-sensitive workloads and zk-v3z when output quality is the priority.

---

API Base URL

https://api.zk0x.app/v1

All endpoints are versioned under /v1. Responses are JSON-encoded.

---

Authentication

All requests require an API key passed via the Authorization header:

Authorization: Bearer ***

API keys are provisioned through the ZKA Model dashboard. Each key is scoped to a project and can be rotated or revoked independently.

Key Type	Description
Production	Full model access, production rate limits
Development	Full model access, reduced rate limits

Security: Never expose API keys in client-side code, public repositories, or logs.

---

Endpoints

Chat Completions

POST /v1/chat/completions

Generate a model response from a conversation.

Request Body:

{
  "model": "zk-v3z",
  "messages": [
    { "role": "system", "content": "You are a helpful assistant." },
    { "role": "user", "content": "Explain zero-knowledge proofs in one paragraph." }
  ],
  "temperature": 0.7,
  "max_tokens": 1024,
  "stream": false
}

Parameter	Type	Required	Default	Description
model	string	Yes	—	Model ID (zk-v3z, zk-v2b, or zk-v1b)
messages	array	Yes	—	List of message objects (role + content)
temperature	float	No	1.0	Sampling temperature (0.0–2.0)
max_tokens	integer	No	Model default	Maximum tokens in the response
stream	boolean	No	false	Enable streaming (see below)
top_p	float	No	1.0	Nucleus sampling threshold
stop	string | array	No	null	Stop sequences

Response (non-streaming):

{
  "id": "chatcmpl-abc123",
  "object": "chat.completion",
  "model": "zk-v3z",
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "..." },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 42,
    "completion_tokens": 128,
    "total_tokens": 170
  }
}

---

List Models

GET /v1/models

Returns the list of models available to the authenticated account.

Response:

{
  "data": [
    { "id": "zk-v3z", "object": "model", "created": 1700000000 },
    { "id": "zk-v2b", "object": "model", "created": 1700000000 },
    { "id": "zk-v1b", "object": "model", "created": 1700000000 }
  ]
}

---

Health Check

GET /v1/health

Returns service status. No authentication required.

---

Streaming

Set "stream": true in your request body to receive tokens as they are generated via Server-Sent Events (SSE).

Response format (streaming):

data: {"id":"chatcmpl-abc123","object":"chat.completion.chunk","choices":[{"delta":{"content":"Hello"},"index":0}]}

data: {"id":"chatcmpl-abc123","object":"chat.completion.chunk","choices":[{"delta":{"content":" world"},"index":0}]}

data: {"id":"chatcmpl-abc123","object":"chat.completion.chunk","choices":[{"delta":{},"finish_reason":"stop","index":0}]}

data: [DONE]

Each data: line contains a JSON object with a delta field containing incremental content. The stream terminates with data: [DONE].

---

Rate Limits

Rate limits are enforced per API key and vary by key type and model.

Key Type	Requests / min	Tokens / min
Production	60	150,000
Development	10	10,000

When a rate limit is exceeded, the API returns 429 Too Many Requests with a Retry-After header indicating when to retry.

Rate limit headers included in every response:

X-RateLimit-Remaining: 42
X-RateLimit-Reset: 1700000060

---

Error Codes

Code	Status	Description
invalid_api_key	401	Missing or invalid API key
model_not_found	404	Requested model does not exist or is not available
invalid_request	400	Malformed request body or missing required parameters
rate_limit_exceeded	429	Rate limit hit — check Retry-After header
internal_error	500	Server-side error — retry with exponential backoff

---

Quickstart

curl https://api.zk0x.app/v1/chat/completions \
  -H "Authorization: Bearer *** \
  -H "Content-Type: application/json" \
  -d '{
    "model": "zk-v2b",
    "messages": [
      { "role": "user", "content": "What is ZK-SNARK?" }
    ]
  }'

---

SDKs & Integration

ZKA Model uses an OpenAI-compatible API format. You can integrate with existing OpenAI SDK clients by changing the base URL:

from openai import OpenAI

client = OpenAI(
    base_url="https://api.zk0x.app/v1",
    api_key="YOUR_API_KEY"
)

response = client.chat.completions.create(
    model="zk-v3z",
    messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)

---

Support

• GitHub: github.com/zeroknowledge0x
• Twitter: @zeroknowledge0x
• Telegram: t.me/zeroknowledge0x

---

ZKA Model — Fast, reliable AI inference.