Conifer SDK

The Conifer SDK provides developers with a unified interface for cloud, local, and Bring Your Own Key (BYOK) inference, ensuring transparency through exact cost receipts for each API call. It is available in both TypeScript and Python, alongside a Machine Communications Protocol (MCP) server to support tools that may not directly communicate with OpenAI or other similar services.

Features

• Single API Key: Utilize one API key across all major models, simplifying configuration and usage.
• Cost Transparency: Each API call returns its exact cost, down to the nanodollar, allowing better financial control and budgeting for projects.
• Model Selection: The SDK automatically routes queries to the most appropriate model, optimizing response times based on cost and efficiency.

Example Usage

TypeScript

import { Conifer, textOf } from "conifer-sdk";

const conifer = new Conifer();
const answer = await conifer.chat({
  model: "claude-haiku-4-5",
  messages: [{ role: "user", content: "three names for a build cache" }],
  maxTokens: 200,
  maxCostNanoUsd: 5_000_000, // restrict to max spend of $0.005
});

console.log(textOf(answer));
console.log(answer.receipt.costUsd);  // Exact cost of this API call

Python

from conifer_sdk import Conifer, ChatRequest

conifer = Conifer()
answer = conifer.chat(ChatRequest(
    model="claude-haiku-4-5",
    messages=[{"role": "user", "content": "three names for a build cache"}],
    max_tokens=200,
    max_cost_nano_usd=5_000_000,
))
print(answer.text, answer.receipt.cost_usd)  # Displays the answer and its cost

Detailed Cost Management

The SDK provides extensive details on costs beyond just the response headers, ensuring comprehensive visibility into spending. This empowers teams to analyze usage effectively and make informed decisions.

Compatibility

The Conifer SDK seamlessly interacts with existing tools and libraries, allowing easy migration and integration with minimal changes required. It supports various models through the OpenAI and Anthropic APIs, ensuring that developers can leverage a wide array of capabilities without needing to adjust their processes significantly.

Deferred Processing

For non-interactive tasks, the SDK allows jobs to be deferred for later execution, providing flexibility in handling high-demand tasks that require asynchronous processing.

By using the Conifer SDK, developers gain a powerful tool that enhances access to AI models while ensuring clarity in operational costs.