{"rewrite":{"id":"r_a12a8e0065ba31bfa0d4e4c1","clusterId":"c_f18d884a4022672d18ef54e4","slug":"liquid-ai-releases-lfm2-5-24b-a2b-for-local-agent-workloads","model":"deepseek-v4-1-flash","headline":"Liquid AI Releases LFM2.5-24B-A2B for Local Agent Workloads","summary":"Liquid AI released LFM2.5-24B-A2B, a 24 billion parameter mixture-of-experts model with 245 million active parameters, pretrained on 33 trillion tokens and further trained for agent workflows. The company reports it scored higher than Qwen3.6-35B-A3B, Gemma-4-26B-A4B-it, and Nemotron-3.5-Lightning (30B-A3B) on agent benchmarks including AutomationBench and BFCLv4. Inference support already exists in llama.cpp, MLX, vLLM, SGLang, and ONNX, and the model is downloadable from Hugging Face as an open model.","whyItMatters":"Liquid AI is positioning a smaller-parameter open model against larger Google and Alibaba models on agent tasks specifically, and the tooling support arriving alongside the weights means the comparison can be checked locally rather than taken on faith.","webCardHtml":"\u003cp\u003eThe model runs the Q4_K_M quantization at 184 tokens per second on a MacBook Pro with M5 Max, using 15GB of memory, and at 114 tokens per second on a PC with Ryzen AI Max+ 395. Liquid AI also published a demo in which Hermes Agent launches on a MacBook Pro with M5 Max and LFM2.5-24B-A2B carries out the task \u0026#39;create a briefing about AI models that are trending on Hugging Face.\u0026#39; The weights are open and available on Hugging Face.\u003c/p\u003e","blueskyPost":"LFM2.5-24B-A2B runs on 245 million active parameters per token, so agent workloads can run on local hardware instead of a server. Liquid AI compares it against models many times its active size.","twitterPost":"LFM2.5-24B-A2B activates 245 million parameters per token. Liquid AI is targeting local agent workloads where weights, not benchmarks, decide deployment.","threadsPost":"LFM2.5-24B-A2B already has inference support in llama.cpp, MLX, vLLM, SGLang, and ONNX. Liquid AI's release is aimed at local agent workloads, and the 245 million active parameters are what make running this on a laptop plausible rather than theoretical.","newsletterBlurb":"Liquid AI has released LFM2.5-24B-A2B, a 24 billion parameter mixture-of-experts model with 245 million active parameters, pretrained on 33 trillion tokens and given additional agent-focused training. Liquid AI reports it outscored larger Google and Alibaba models on agent benchmarks including AutomationBench and BFCLv4. Inference support is already in llama.cpp, MLX, vLLM, SGLang, and ONNX, and the open weights are on Hugging Face.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20261014-lfm2-5-24b-a2b/\",\"title\":\"LFM2.5-24B-A2B debuts, outperforms Gemma 4 26B-A4B and Qwen3.6 35B-A3B as a local agent\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":4917,"outputTokens":774,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1791967029,"createdAt":"2026-10-14T08:31:59.000Z","publishedAt":"2026-10-14T08:36:06.000Z","updatedAt":"2026-10-14T08:36:06.000Z"},"cluster":{"id":"c_f18d884a4022672d18ef54e4","canonicalTitle":"「LFM2.5-24B-A2B」が登場、Gemma 4 26B-A4BやQwen3.6 35B-A3Bより高性能でローカルエージェントとして能力を発揮","representativeArticleId":"a_b3123783b8d972f51155b376","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":true,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[\"LFM2.5-24B-A2B\"],\"studios\":[\"Liquid AI\"],\"people\":[],\"type\":\"announcement\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-10-14T07:30:00.000Z","lastSeenAt":"2026-10-14T07:30:00.000Z","updatedAt":"2026-10-14T08:36:06.000Z"},"attribution":[{"source":"GIGAZINE","url":"https://gigazine.net/news/20261014-lfm2-5-24b-a2b/","title":"「LFM2.5-24B-A2B」が登場、Gemma 4 26B-A4BやQwen3.6 35B-A3Bより高性能でローカルエージェントとして能力を発揮"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":["LFM2.5-24B-A2B"],"studios":["Liquid AI"],"people":[],"type":"announcement","domain":"other","is_roundup":false},"keyFacts":null}