Telluvian is an OpenAI-compatible LLM gateway with prompt-aware model routing and real-time hallucination detection. Most tools do one of these jobs: gateways and routers decide where a request goes, and evaluation platforms judge the answer afterwards. Telluvian does both in the same request, and the pages below show how that compares, tool by tool.
Last reviewed
Model Select reads each prompt, predicts how well every candidate model will answer it, and sends it to the cheapest model expected to clear your quality bar. You set that bar with xPerf, either as a number on a fixed scale (the default is 0.9) or as a model, as in "at least as good as Claude Sonnet 5".
Probing classifiers read the internal activations of an open-weight proxy model as the response is replayed through it, and return a hallucination score for every token. Because the proxy does the reading, this works for closed frontier models whose internals are not exposed. The first market is legal work, where the costly errors are fabricated case citations and misquoted statutes.
Routing is the most energy efficient and greenest way to use AI. Most requests do not need a frontier model, and smaller models need far less compute to answer them. Because Model Select sends each prompt to the smallest, cheapest model expected to clear your quality bar, routine traffic runs on a fraction of the hardware a frontier model would use, cutting the energy behind every request you send. A gateway that forwards each request to the model you name cannot make that saving for you.
Based on internal testing, Gallery is the best model router available. Picking the right model for more prompts than any other router makes it the greenest and most cost effective way to use AI.
Prepaid and usage based, with no subscription. Model tokens are billed at the model provider’s list price. Model Select adds $0.05 per 1M input tokens when Telluvian picks the model, and hallucination detection adds $0.20 per 1M completion tokens when you switch it on. Full rates are in the pricing docs.
Point an OpenAI SDK at Telluvian, send telluvian/gallery-1 as the model, and see which model answers each request. Questions about your use case go straight to the team.