Code Examples

Copy-paste patterns for Elastic (chat, tools, embeddings). For Dedicated, change the URL to /api/d/{endpoint-id}/chat/completions and use the correct auth. Prefer catalog display names (e.g. claude-sonnet-5) for the model field.

cURL (Streaming)

BASH

curl -N https://hypervize.tech/api/chat/completions \
  -H "Authorization: Bearer $HVZ_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mistral-large-3",
    "messages": [{"role": "user", "content": "Explain RAG in 3 bullet points"}],
    "max_tokens": 300,
    "stream": true
  }'

Python (OpenAI SDK — Recommended)

PYTHON

from openai import OpenAI

client = OpenAI(
    base_url="https://hypervize.tech/api",
    api_key="hvz_live_..."
)

stream = client.chat.completions.create(
    model="claude-sonnet-5",
    messages=[{"role": "user", "content": "Write a haiku about GPUs"}],
    max_tokens=150,
    stream=True,
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

JavaScript / TypeScript (fetch)

const response = await fetch("https://hypervize.tech/api/chat/completions", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.HVZ_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "llama-3-3-70b-instruct",
    messages: [{ role: "user", content: "Hello" }],
    stream: true,
  }),
});

const reader = response.body?.getReader();
const decoder = new TextDecoder();

while (true) {
  const { value, done } = await reader!.read();
  if (done) break;
  const chunk = decoder.decode(value);
  // Parse SSE lines starting with "data: "
  console.log(chunk);
}

Tool Calling Examples

Plain OpenAI tools (client executes) – cURL

BASH

curl -N https://hypervize.tech/api/chat/completions \
  -H "Authorization: Bearer $HVZ_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5",
    "messages": [{"role":"user","content":"What is the weather in Nashville?"}],
    "tools": [{
      "type": "function",
      "function": {
        "name": "get_weather",
        "description": "Get current weather for a city",
        "parameters": {
          "type": "object",
          "properties": {"city": {"type": "string"}},
          "required": ["city"]
        }
      }
    }],
    "stream": true
  }'

Webhook tool (server-managed)

Note: Server-managed tools require "stream": true.

JSON

{
  "tools": [{
    "type": "function",
    "function": { "name": "my_internal_tool", "description": "...", "parameters": {...} },
    "webhook": {
      "url": "https://myapp.com/api/ai-tools/internal",
      "key": "my-secret-key-123",
      "timeout_seconds": 25
    }
  }]
}

Hypervize will call your endpoint with a signed payload and drive the loop.

Using Built-in Platform Tools

Platform tools you can enable in Alexandria (or via /api/tools):

Product	Callable names	Notes
Athena	`athena`	Time/date and basic math
Vesper	`vesper`	Web search and page extraction
Herald	`herald`	Email (explicit user permission required); optional `attachment_file_id` from Ledger
Pandora	`pandora_*`	Google Workspace / Gmail suite
Iris	`iris`	Image generation from text
Ledger	`ledger`	Create/read/convert files (CSV, PDF, JSON, …); Library → Files
Chronos	—	Product flag only — enables Chat Schedule and Chronos API jobs; not injected as a model tool

BASH

curl -N https://hypervize.tech/api/chat/completions \
  -H "Authorization: Bearer $HVZ_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mistral-large-3",
    "messages": [{"role": "user", "content": "What time is it and what is 15 * 7?"}],
    "stream": true
  }'

When Athena (or other platform tools) is enabled for your account, those tools are available on streaming requests. Hypervize executes them server-side and continues the turn.

See the full Platform Tools reference for parameters and pricing.

Vercel AI SDK with custom tools

See the full example in the Tool Calling guide.

LangChain (Python)

PYTHON

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="https://hypervize.tech/api",
    api_key="hvz_live_...",
    model="qwen3-32b",
)

response = llm.invoke("Explain why streaming matters for LLMs")
print(response.content)

LlamaIndex

LlamaIndex works via the same OpenAI-compatible base URL. Set:

PYTHON

from llama_index.llms.openai import OpenAI

llm = OpenAI(
    api_base="https://hypervize.tech/api",
    api_key="hvz_live_...",
    model="nemotron-3-super-120b",
)

Embeddings

Python (OpenAI SDK)

PYTHON

from openai import OpenAI

client = OpenAI(
    base_url="https://hypervize.tech/api",
    api_key="hvz_live_..."
)

response = client.embeddings.create(
    model="cohere.embed-v4",
    input="The quick brown fox jumps over the lazy dog."
)

print(response.data[0].embedding[:5])  # first 5 dimensions

cURL

BASH

curl https://hypervize.tech/api/embeddings \
  -H "Authorization: Bearer $HVZ_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cohere.embed-v4",
    "input": "The quick brown fox jumps over the lazy dog."
  }'

JavaScript / TypeScript

const response = await fetch("https://hypervize.tech/api/embeddings", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.HVZ_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "cohere.embed-v4",
    input: "The quick brown fox jumps over the lazy dog.",
  }),
});

const data = await response.json();
console.log(data.data[0].embedding.length); // dimension count

Notes for Production Clients

Always set a reasonable timeout / read_timeout.
Implement retry logic with exponential backoff on 429 / 5xx. Some Elastic models may already try an alternate catalog model under provider rate limits before you see an error; still handle 429/5xx when capacity is exhausted.
Parse usage from the final chunk for cost tracking.
Prefer the official OpenAI SDK when possible — it handles SSE edge cases well.

Scheduled tools (Chronos)

To run ordered platform tools and optional model steps on a schedule (API key, full JavaScript client), see Chronos. Chronos is enabled as a product toggle; it is not a model-callable function.

More Examples

Need an example for a specific framework (Vercel AI SDK, AutoGen, CrewAI, etc.)? Let us know — we are expanding this section.