Route LLM calls across GPT, Claude & Gemini via one API

RouterBase Model Gateway is a skill for using routerbase.com's single OpenAI-compatible API surface to route requests across GPT, Claude, Gemini, image, video, audio, and embedding workloads. It helps agents migrate existing OpenAI SDK calls to RouterBase, document model-selection tradeoffs, and produce safe implementation snippets that never expose credentials. Because RouterBase model availability, pricing, and provider capabilities can change, the skill treats its own examples as starting points and calls for verifying current catalog data before any production recommendation.

Use it when migrating an OpenAI-compatible client to RouterBase by changing the base URL and model ID, when selecting primary and fallback models for chat, reasoning, vision, media generation, audio, or embeddings, when debugging RouterBase request setup (headers, environment variables, streaming, tool calls, JSON mode, multimodal payloads), or when documenting an internal model-routing plan that balances cost, latency, quality, and provider redundancy. It does not replace RouterBase account configuration, live model catalog checks, or production observability, and high-stakes outputs still need human review and domain-specific evaluation - some model features are provider-specific and must be tested against the exact selected model.

Input is an OpenAI-compatible request (messages, model ID, and any streaming/tool-calling/JSON-mode options) sent with a RouterBase API key kept server-side in an environment variable such as ROUTERBASE_API_KEY. Output is the same OpenAI-compatible response shape from whichever underlying provider RouterBase routed the request to, for example:

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["ROUTERBASE_API_KEY"],
    base_url="https://routerbase.com/v1",
)

response = client.chat.completions.create(
    model="google/gemini-2.5-flash",
    messages=[{"role": "user", "content": "Write one sentence about model routing."}],
)

print(response.choices[0].message.content)

The skill's recommended migration checklist is: change the base URL to https://routerbase.com/v1, read the API key from server-side environment configuration, replace the model name with a matching RouterBase model ID, preserve standard OpenAI request fields unless RouterBase documentation says otherwise, and run one minimal smoke test before shipping.

It sits on top of the standard OpenAI SDK (Python and JavaScript) via a base_url override, and can be queried directly for its live model catalog, for example a chat-task listing at routerbase.com/api/v1/models?task=chat with a Bearer token. It recommends explicit application-level fallback chains across providers, for example trying anthropic/claude-sonnet-4-6 before falling back to google/gemini-2.5-flash, unless the account already has a smart-routing policy configured, retrying only transient network errors, timeouts, rate limits, and server errors while failing fast on authentication, invalid-model, validation, and policy-refusal errors. Related skills: api-analyzer for validating a single API request shape, and langfuse for production LLM observability, tracing, and evaluation.

Engineers migrating an OpenAI-compatible integration to a multi-provider gateway, or documenting a model-routing plan that weighs cost, latency, quality, and provider redundancy - anyone who needs primary and fallback model selection with credentials kept strictly server-side, never pasted, logged, committed, screenshotted, or exposed in client-side JavaScript, mobile apps, or public repositories.