Developer Tools

Integrations & Developer Tools

OpenAI-compatible API, first-party SDKs for Python and TypeScript, a CLI tool, and a VS Code extension. Everything you need to migrate from OpenAI and start building in minutes.

OpenAI-Compatible API

Alveare's inference API follows the OpenAI chat completions format. If your application currently calls the OpenAI API, you can switch to Alveare by changing two values: the base URL and the API key. Your request and response formats remain identical.

This is not a partial compatibility layer. We support the full chat completions interface including system messages, multi-turn conversations, streaming responses, temperature and top-p parameters, stop sequences, max tokens, and JSON mode. The goal is that your existing code works without modification.

OpenAI format — works with Alveare unchanged
# Before: OpenAI import openai client = openai.OpenAI( api_key="sk-...", base_url="https://api.openai.com/v1" ) # After: Alveare (change 2 lines) client = openai.OpenAI( api_key="alv_live_abc123...", base_url="https://api.alveare.ai/v1" ) # Everything else stays the same response = client.chat.completions.create( model="classify", # specialist name as model messages=[ {"role": "user", "content": "Classify this ticket: My payment failed..."} ], temperature=0.1, max_tokens=50 ) print(response.choices[0].message.content) # "billing"

Python SDK

The Alveare Python SDK provides a typed, ergonomic interface on top of the API. It handles authentication, retries, streaming, and error handling. Install it with pip and start making requests in three lines.

pip install alveare
from alveare import Alveare client = Alveare(api_key="alv_live_abc123...") # Simple inference result = client.infer( specialist="summarise", prompt="Summarise this quarterly earnings report...", max_tokens=256 ) print(result.text) # "Revenue grew 23% year-over-year driven by..." print(result.tokens_used) # 142 print(result.latency_ms) # 312 # Streaming for chunk in client.infer_stream( specialist="chat", prompt="Explain cognitive hive architecture" ): print(chunk.text, end="") # Batch inference results = client.infer_batch( specialist="classify", prompts=[ "My payment didn't go through", "How do I reset my password?", "Your product is great, thanks!" ] ) for r in results: print(r.text) # "billing" # "account" # "feedback"

The Python SDK supports Python 3.9+ and has zero required dependencies beyond the standard library. Optional dependencies for async support (httpx) and streaming (sseclient) are installed with pip install alveare[async,stream].

TypeScript SDK

The TypeScript SDK provides full type safety, native Promise support, and works in Node.js, Deno, and Bun. It exports both ESM and CommonJS builds. Install with npm, yarn, or pnpm.

npm install @alveare-ai/sdk
import { Alveare } from '@alveare-ai/sdk'; const client = new Alveare({ apiKey: 'alv_live_abc123...', }); // Simple inference const result = await client.infer({ specialist: 'extract', prompt: 'Extract all company names from this press release...', outputFormat: 'json', }); console.log(result.text); // '{"companies": ["Acme Corp", "Globex", "Initech"]}' console.log(result.tokensUsed); // 89 // Streaming with async iterator for await (const chunk of client.inferStream({ specialist: 'chat', prompt: 'Explain the benefits of private inference', })) { process.stdout.write(chunk.text); } // Type-safe specialist configuration const specialists = await client.listSpecialists(); // TypeScript knows the shape of each specialist for (const s of specialists) { console.log(`${s.name}: ${s.requestCount} requests`); }

CLI Tool

The Alveare CLI lets you manage specialists, make inference requests, and monitor usage from the terminal. It is useful for testing prompts, scripting batch jobs, and CI/CD integration.

Terminal
# Install brew install alveare/tap/alveare # or: curl -fsSL https://get.alveare.ai | sh # Authenticate alveare auth login API key: alv_live_abc123... Authenticated as acme-corp # Make an inference request alveare infer --specialist classify --prompt "My payment failed" billing (142ms, 15 tokens) # Pipe input from a file cat report.txt | alveare infer --specialist summarise --max-tokens 256 # List specialists alveare specialists list NAME MODEL REQUESTS/MO AVG LATENCY classify mistral-7b 42,391 89ms summarise mistral-7b 18,204 290ms extract mistral-7b 31,847 165ms # Check usage alveare usage Plan: Professional Requests: 92,442 / 500,000 (18.5%) Billing period: Mar 1 - Mar 31 # Create a new specialist alveare specialists create \ --name sentiment \ --system-prompt "Classify the sentiment as positive, negative, or neutral." \ --temperature 0.0 \ --max-tokens 10

VS Code Extension

The Alveare VS Code extension brings inference testing into your editor. Test prompts, preview specialist outputs, and monitor your hive without leaving your development environment.

Inline prompt testing

Select text in any file, right-click, and send it to a specialist. The response appears in a side panel with latency and token count.

Specialist browser

View all your configured specialists in the sidebar. Edit system prompts, generation parameters, and validation rules directly in VS Code.

Usage dashboard

A status bar item shows your current request count and remaining allocation. Click it to open a detailed usage view with per-specialist breakdowns.

Response diff view

Compare responses from different specialists or different parameter configurations side by side. Useful for prompt engineering and quality testing.

Install from the VS Code Marketplace: search for "Alveare" or run ext install alveare.alveare-vscode from the command palette.

Webhook Support

Alveare sends webhook notifications for billing and usage events. Configure webhook endpoints in the dashboard and receive real-time notifications when usage thresholds are hit, billing events occur, or specialists change status.

Supported events

Webhook payload example
{ "event": "usage.threshold", "timestamp": "2026-03-17T14:30:00Z", "data": { "plan": "professional", "current_requests": 375000, "limit": 500000, "percentage": 75, "threshold_configured": 75, "days_remaining": 14 }, "signature": "sha256=a1b2c3d4..." }

All webhook payloads include an HMAC-SHA256 signature for verification. The signing secret is generated when you configure the webhook endpoint and can be rotated at any time.

Migrate from OpenAI in 15 Minutes

This is a summary of the migration process. The full step-by-step guide is available in the documentation.

Sign up and get your API key

Create an Alveare account and choose a plan. Your hive is provisioned automatically and you receive an API key within 60 seconds.

~2 minutes

Configure your specialists

Define specialists that match your current OpenAI use cases. For each specialist, set a system prompt, temperature, max tokens, and output format. Use the dashboard or CLI.

~5 minutes

Update your base URL and API key

Change api.openai.com to api.alveare.ai and replace your OpenAI key with your Alveare key. If you use the OpenAI Python or Node SDK, change the base_url parameter.

~2 minutes

Run your test suite

Execute your existing integration tests. Because Alveare uses the same request/response format, most tests pass without modification. Adjust any tests that check for specific model names.

~3 minutes

Deploy and monitor

Push the URL change to production. Monitor the usage dashboard and specialist latency for the first hour. If latency or quality does not meet expectations, adjust specialist parameters without redeploying.

~3 minutes

Total migration time for a typical integration: under 15 minutes. No code rewrite. No SDK change. No downtime. If you run into issues, our support team is available to help with the migration.

Start building

Get your API key and make your first request in under 5 minutes. Full SDK documentation included.

Get Started Free