PulseAugur / Brief
EN
LIVE 21:50:41

Brief

last 24h
[2/2] 222 sources

Multi-source AI news clustered, deduplicated, and scored 0–100 across authority, cluster strength, headline signal, and time decay.

  1. SIGNIFICANT · Hacker News — AI stories ≥50 points English(EN) · · [2 sources]

    Liquid AI reveals 8B-A1B MoE trained on 38T

    Liquid AI has released LFM2.5-8B-A1B, an 8-billion parameter Mixture-of-Experts model designed for efficient on-device tool calling. This new version significantly expands the context window to 128,000 tokens and scales pretraining to 38 trillion tokens, improving its ability to process long documents and reason over extended tasks. The model also features an expanded vocabulary for better non-Latin language tokenization and is optimized for high throughput on consumer hardware, supporting various inference engines. AI

    IMPACT Enables more capable on-device AI applications by improving reasoning and context handling for consumer hardware.

  2. FRONTIER RELEASE · Hugging Face Trending Models Français(FR) · · [5 sources]

    LiquidAI/LFM2.5-8B-A1B-GGUF

    Liquid AI has released LFM2.5-8B-A1B, a new on-device Mixture-of-Experts (MoE) model designed for complex tasks and tool chaining. This model features 8.3 billion total parameters but activates only 1.5 billion per token, enabling efficient performance on consumer hardware. Key upgrades include an expanded 131,072 token context window, extensive pre-training on 38 trillion tokens, and advanced reinforcement learning to reduce hallucinations and improve reasoning. AI

    IMPACT Enables more capable on-device AI applications with competitive performance and reduced hallucination rates.