According to Beating Monitor, Liquid AI has released LFM2.5-2.6B and opened its model weights. This on-device agent model has only 2.6 billion parameters and, after quantization, occupies approximately 2.5 GB of memory, achieving a generation speed of about 30 tokens per second on mobile devices. In official evaluations, it outperforms Qwen3.5-9B in three benchmarks: tool calling, multi-turn instruction following, and structured output. Qwen3.5-9B, with 9.7 billion parameters, is nearly four times larger. However, Qwen3.5-9B still leads in code generation and certain complex agent tasks.
Liquid AI launches a 2.6B-parameter on-device model that outperforms Qwen3.5-9B in three tests.
MarsBitShare
Summary
Liquid AI has released its LFM2.5-2.6B model, a notable on-chain development in AI and crypto news. This open-weight model features 2.6 billion parameters and consumes approximately 2.5GB of memory after quantization. It generates tokens at a rate of 30 per second on mobile devices. In evaluations, it outperformed Qwen3.5-9B in tool calling, multi-turn instruction following, and structured output. However, Qwen3.5-9B remains superior in code generation and complex agent tasks.
Source:Show original
Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information.
Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.