anyask

Call an LLM provider with a single text prompt, get its raw response back — one function, 18 vendors, zero routing/fallback magic.

import anyask

response = anyask.ask(
    "Say hello in one word.",
    provider="anthropic",
    model="claude-haiku-4-5-20251001",
    api_key="sk-...",  # or set ANTHROPIC_API_KEY
)

print(response.content)   # "Hello"
print(response.usage)     # TokenUsage(prompt_tokens=..., completion_tokens=..., total_tokens=...)

Features

• Lightweight installation — a bare pip install anyask pulls in exactly one dependency (PyYAML); each provider you actually use adds only that provider's own SDK, never the other 17. Cheap to bake into a container image, fast to install in CI.
• One function, 18 vendors — ask()/ask_async() take the same provider=/model= signature everywhere; swap providers by changing one string.
• No routing/fallback magic — provider and model are always explicit; anyask never silently swaps one vendor for another.
• Real reasoning support, not simulated — reasoning=True enables each provider's own actual extended-thinking mechanism where one exists.
• A small built-in CLI — anyask check/anyask ready answer "is this provider's SDK installed and its credentials set," with no network call.

Installation

pip install anyask                 # minimal install — no provider usable yet
pip install anyask[anthropic]      # + Anthropic only
pip install anyask[openai,gemini]  # + multiple providers
pip install anyask[all]            # + every provider SDK

Verify:

anyask version                     # installed version
anyask ready                       # sweep every provider's SDK/credential status
anyask check anthropic             # check one provider specifically, exit 0/1

Supported providers

openai, anthropic, gemini, vertexai, azure, mistral, cohere, bedrock, oci, and the nine OpenAI-compatible vendors: deepseek, groq, xai, together, fireworks, cerebras, perplexity, openrouter, moonshot.

What is anyask

A single-responsibility package for one job: call an LLM provider, get its raw response back. prompt is always a plain string and AskResponse.content is always a plain string — no multi-turn history, no system prompts, no multi-modal input/output, no streaming. That narrow contract is deliberate, not a missing feature — see SPEC.md for the full design and SPEC.md's out-of-scope list for exactly what's excluded and why.

Why do you need anyask

If you're calling more than one LLM vendor, you're already facing 18 SDKs with 18 different client shapes, auth patterns, and response formats. anyask is the thin, uniform layer underneath: one call signature, one normalized response shape (content/usage/finish_reason/raw), and nothing else — no opinions about retries, fallback, cost tracking, or agent orchestration. Build those on top if you need them, or just use anyask directly when you don't. It's also the option to reach for when install footprint actually matters — CI pipelines, container images, or a library that can't afford to drag in SDKs its own users never touch.

Documentation

Full docs, including the Quickstart, Providers reference, CLI Guide, and generated API Reference:

• Latest: https://geeks-trident-llc.github.io/anyask/latest/
• All versions: https://geeks-trident-llc.github.io/anyask/

License

MIT