Ant Group's inclusionAI has launched Ling-2.6-1T, a new trillion-parameter AI model designed for precise task execution. This model, part of the Ling series, utilizes a "Fast-Thinking" mechanism, distinguishing itself from models focused on long-form reasoning. Ling-2.6-1T is touted as a significant upgrade over its predecessor, Ling-1T, with intelligence levels nearing GPT-5.4 in non-reasoning tasks while maintaining token efficiency as a core metric. The model is tailored for agent-era execution tasks such as code modification, tool invocation, and complex instruction following. It supports a 262.1k token context window and can output up to 32.8k tokens. Currently available via a free API on the OpenRouter page, Ling-2.6-1T is set to be open-sourced soon, according to the official team.