Description
Agnes AI Agnes 3.0 Flash — the vendor's new-generation language model, built for agent coding and tool-driven task execution, with a 512K input context and a 65,536-token output limit. Takes text and image-URL input and returns text, and supports streaming, function calling and multi-step tool orchestration, and long-task instruction following. Reachable on the OpenAI-compatible chat endpoint, the Responses endpoint and the Anthropic-compatible Messages endpoint. On the chat endpoint, Thinking mode is switched on by setting chat_template_kwargs.enable_thinking to true; measured 2026-09-21, requests without the field returned no reasoning tokens. With Thinking on, reasoning tokens count toward max_tokens, so size it for the reasoning as well as the answer.
Capabilities
- Input modalities
- text, image
- Output modalities
- text
- Tasks
- chat, reasoning, vision
API endpoints
POST /v1/chat/completions(openai)POST /v1/responses(openai-response)POST /v1/messages(anthropic)