unifiedai-sdk 1.0.3

Unified SDK for multi-provider LLM comparison (Cerebras, AWS Bedrock) with OpenAI-compatible interface.

Unified AI SDK

OpenAI-compatible Python SDK that unifies multiple providers (Cerebras, AWS Bedrock) with Solo and Comparison modes and built-in telemetry.

Highlights

• OpenAI-like API surface: UnifiedAI().chat.completions.create(...)
• Providers via adapters: Cerebras, Bedrock (extensible)
• Modes: Solo and side-by-side Comparison
• Telemetry: total round-trip time, time-to-first-byte, tokens, costs (extensible)

Install

pip install unifiedai

Install from GitHub (optional)

pip install git+https://github.com/<your-org-or-user>/<your-repo>.git#subdirectory=cerebras

Note: The subdirectory=cerebras flag installs the package defined under cerebras/pyproject.toml.

Quickstart

from unifiedai import UnifiedAI

client = UnifiedAI(provider="cerebras", model="llama3")
resp = client.chat.completions.create(messages=[{"role": "user", "content": "Hello"}])
print(resp.choices[0].message["content"])  # OpenAI-like shape

Project Structure

• src/unifiedai/: SDK implementation
• tests/: unit tests
• examples/: usage examples
• apps/chat/: demo chat UI

License

MIT