Colibri language model runs 744B parameters on laptop with 25GB RAM

By PulseAugur Editorial · [1 sources] · 2026-07-11 16:01

A new language model called Colibri, based on GLM-5.2, has been developed to run on a laptop with only 25GB of RAM. This model, which has 744 billion parameters, operates in pure C with no external dependencies. The efficiency is achieved by activating only approximately 40 billion parameters at a time. AI

IMPACT Enables running large language models on consumer hardware, potentially democratizing access and use.

RANK_REASON The item describes a new language model release with technical details about its architecture and resource requirements. [lever_c_demoted from research: ic=1 ai=1.0]

Read on Mastodon — fosstodon.org →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Colibri language model runs 744B parameters on laptop with 25GB RAM

COVERAGE [1]

Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] · 2026-07-11 16:01

RT @chenzeling4: 744B parameters. On a laptop. With 25GB RAM. Colibri runs GLM-5.2 (744B MoE) in pure C with zero dependencies. The trick: only ~40B params acti

RT @chenzeling4: 744B parameters. On a laptop. With 25GB RAM. Colibri runs GLM-5.2 (744B MoE) in pure C with zero dependencies. The trick: only ~40B params activate per token, so it keeps the dense part resident and streams experts from disk on demand. A single 2,400-line C file.…

LINKS github.com/…/colibri

COVERAGE [1]

RT @chenzeling4: 744B parameters. On a laptop. With 25GB RAM. Colibri runs GLM-5.2 (744B MoE) in pure C with zero dependencies. The trick: only ~40B params acti

RELATED ENTITIES

RELATED TOPICS