{"version":"1.0","type":"card","id":"28932273-e226-49ad-8b71-43a3a7cd7137","url":"https://stacklist.com/card/28932273-e226-49ad-8b71-43a3a7cd7137","title":"UC Berkeley Open-Sources FreeToken for LLM Inference","source_url":"https://x.com/akshay_pachaar/status/2091150763418620133?s=12","note":"UC Berkeley has released FreeToken, a tool that enables 2–4x faster local LLM inference compared to Ollama. The performance results showcase impressive speeds for various models on different GPU configurations.","image":{"url":"https://ucarecdn.com/f69f43ca-37ec-443a-8169-52809e317413/","alt":"UC Berkeley Open-Sources FreeToken for LLM Inference","width":1200,"height":480},"stack":{"id":"ed01fcda-d7b5-4a4b-bdfa-bc07c9df5228","title":"Local AI & GPUs","url":"https://stacklist.com/c/technology/stack/ed01fcda-d7b5-4a4b-bdfa-bc07c9df5228"},"created_at":"2026-08-24T02:54:48.312Z","updated_at":null,"aco":{"summary":"FreeToken is an open-source inference engine from UC Berkeley that achieves 2-4x faster local LLM inference than Ollama by intelligently routing Mixture-of-Experts models between GPU and CPU memory. The system dynamically profiles hardware bandwidth and optimizes expert placement, enabling large models like Qwen 35B to run on consumer 8GB GPUs.","tags":["freetoken","llm-inference","mixture-of-experts","gpu-optimization","open-source","local-inference"],"key_entities":[{"name":"FreeToken","type":"technology","confidence":0.99},{"name":"UC Berkeley","type":"organization","confidence":0.99},{"name":"Akshay Pachaar","type":"person","confidence":0.99},{"name":"Qwen3.6-35B","type":"technology","confidence":0.95},{"name":"DeepSeek-V4-Flash","type":"technology","confidence":0.95},{"name":"GLM-5.2","type":"technology","confidence":0.95},{"name":"Ollama","type":"technology","confidence":0.95},{"name":"Mixture-of-Experts","type":"concept","confidence":0.98},{"name":"OpenAI API","type":"technology","confidence":0.9},{"name":"Anthropic API","type":"technology","confidence":0.9}],"classification":"analysis","language":"en","confidence":0.85,"provenance":{"model":"claude-haiku-4-5","tool":"@stacklist/be@0.1.0","confidence":0.85,"timestamp":"2026-08-24T02:54:55.105Z"},"token_counts":{"approximate":893,"cl100k":839},"content_hash":"sha256:6e710b46deab38eb09d4acfaae96c9ca412aec8af0f75ee62c2632622f31173a","acp_version":"0.2","body_available":true,"body_tokens":893,"visibility":"public","agent_accessible":true,"status":"final"},"_links":{"self":"/api/public/card/28932273-e226-49ad-8b71-43a3a7cd7137.json","html":"https://stacklist.com/card/28932273-e226-49ad-8b71-43a3a7cd7137","md":"/api/public/card/28932273-e226-49ad-8b71-43a3a7cd7137.md","stack_json":"/api/public/stack/ed01fcda-d7b5-4a4b-bdfa-bc07c9df5228.json"}}