AionLabs: Aion 3.5 Mini

Aion 3.5 Mini is a text model from AionLabs, listed on api14 under the model id aion-labs/aion-3.5-mini. AionLabs describes it as a multi-model roleplaying and storytelling system built on the GLM family of models, and positions it as the smaller sibling of Aion 3.5. That description matters when you are choosing between the two: the Mini variant is the lighter option in the same line, aimed at the same kind of work rather than at general-purpose chat alone.

The model takes text in and returns text out, so there is no image or audio handling to plan around. Its context window is 262144 tokens, which is large enough to hold long conversations, extended narrative drafts or sizeable reference material in a single request. Output is capped at 32768 tokens. On the request side it supports max_tokens, temperature, top_p and response_format, along with tools and tool_choice for function calling. It also exposes reasoning controls: include_reasoning, reasoning and reasoning_effort, so you can decide whether reasoning content is returned and how much effort the model spends on it.

Those capabilities point to a fairly specific set of uses. Roleplay and interactive storytelling are the stated focus, and the long context suits multi-turn character work where earlier scenes need to stay in view. The same properties fit long-form fiction drafting, branching narrative games and dialogue systems that must keep a consistent voice across many turns. Because tools and structured output via response_format are available, it can also sit inside an application that needs to call functions or return JSON-shaped results while still producing prose, for example a story engine that looks up world state between scenes.

Calling it through api14 works like any other OpenAI-compatible endpoint. Point your client at the api14 base URL, send a chat completions request, and set the model field to aion-labs/aion-3.5-mini. The parameters listed above can be passed in the request body in the usual way: max_tokens, temperature, top_p, response_format, tools and tool_choice, plus the reasoning options include_reasoning, reasoning and reasoning_effort. No special SDK is required if your existing code already targets an OpenAI-compatible API.

The main thing to weigh is scope. Inputs and outputs are text only, so tasks involving images or audio need a different model. The 32768 token output cap also means very long single responses have to be split or continued across calls. And since AionLabs frames this as a roleplaying and storytelling system, a general-purpose model may be a better fit for tasks far outside narrative and conversational work.