PulseAugur
EN
LIVE 09:42:50

Cactus Compute releases Needle 2, a 14MB tool-calling AI model for edge devices

By PulseAugur Editorial · [2 sources] ·

Cactus Compute has launched Needle 2, a compact 45-million-parameter model designed for tool-calling and structured data extraction. This model is notable for its extremely small footprint, shipping as a 14MB binary and requiring only 28MB of RAM to operate, making it suitable for devices without dedicated AI hardware like GPUs or NPUs. Needle 2 achieves impressive speeds, processing up to 500 tokens per second on a Raspberry Pi 5 and offering efficient performance on various mobile and mixed-reality devices. AI

IMPACT Enables sophisticated AI capabilities on low-power, resource-constrained devices, expanding AI applications beyond traditional cloud or high-end hardware.

RANK_REASON Model release from a specialized AI lab focused on efficient inference. [lever_c_demoted from frontier_release: ic=2 ai=1.0]

Read on Mastodon — mastodon.social →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

Cactus Compute releases Needle 2, a 14MB tool-calling AI model for edge devices

COVERAGE [2]

  1. MarkTechPost TIER_1 English(EN) · Michal Sutter ·

    Meet Needle 2: An Open 45M-Parameter Tool-Calling Model That Ships as a 14MB Binary and Runs a Full Session in 28MB of RAM

    <p>Cactus Compute released Needle 2, an open 45M-parameter model for tool calling, device use, and structured extraction. The full model is a single 14MB binary that runs a session in about 28MB of RAM. It leads both Seal-Tools splits while targeting hardware with no GPU and no N…

  2. Mastodon — mastodon.social TIER_1 English(EN) · [email protected] ·

    Cactus Compute has released Needle 2, an open 45M-parameter tool-calling model that runs in a 14MB binary using just 28MB RAM. It targets hardware with no GPU a

    Cactus Compute has released Needle 2, an open 45M-parameter tool-calling model that runs in a 14MB binary using just 28MB RAM. It targets hardware with no GPU and achieves 500 tokens/sec on a Raspberry Pi 5. https://www. marktechpost.com/2026/08/13/ca ctus-compute-needle-2-45m-pa…

RELATED ENTITIES

RELATED TOPICS