According to monitoring by Dongcha Beating, Ant Group's inclusionAI has recently released Ling-2.6-1T, a new trillion-parameter flagship instruction model in the Ling series. It is designed not for long reasoning processes but for precise task execution using a 'Fast-Thinking' mechanism. The company claims that it has a significant upgrade in capabilities compared to Ling-1T, with overall intelligence approaching GPT-5.4 in non-reasoning mode, while producing fewer tokens and prioritizing token efficiency as a core metric. This indicates that it targets execution tasks in the agent era, such as code modification, tool invocation, complex instruction following, and high-frequency production environment calls. The OpenRouter page shows that Ling-2.6-1T currently offers a free API entry, supporting a context of 262.1k tokens and a maximum output of 32.8k tokens. The company has announced that Ling-2.6-1T will soon be open-sourced.
All Comments