Venice
Models published by Venice, available through the AnyRouter API. Each can route across multiple upstream providers for availability and price.
Google Gemma 4 26B A4B Instruct is Google DeepMind's Mixture-of-Experts model with 26B total and 4B active parameters per token. It takes text, image and video input and supports 256K context, function calling and reasoning with configurable thinking modes.
GPT-5.4 Mini brings the core capabilities of GPT-5.4 to a faster, more efficient model for high-throughput workloads. It takes text and image input and supports reasoning, coding, tool use and structured output.
MiniMax M2.7 is a large language model built for autonomous, real-world productivity, with agentic capabilities through multi-agent collaboration. It supports reasoning, function calling and structured output with a 198K context window.
Venice Role Play Uncensored is an uncensored model tuned for creative roleplay, immersive storytelling and open-ended writing. It supports vision input, function calling and structured output, with a 128K context window.
NVIDIA Nemotron 3 Ultra is built for frontier reasoning, orchestration, coding agents and deep research. It supports up to 1M token context, function calling and structured output, and targets complex enterprise workflows.
Hermes 3 Llama 3.1 405B is Nous Research's flagship 405B-parameter model for reasoning, coding and instruction following, built on Llama 3.1. It supports function calling and structured output, which suits agentic workflows.