Daily AI Briefing: October 15, 2026
Google releases Gemini 4 Argon with a 1M token output limit, and Liquid AI launches LFM2.5-24B-A2B, a 2B-active MoE for on-device agents.
Models and Research
Google Releases Gemini 4 Argon API With 1M Token Output Limit
Google DeepMind made Gemini 4 Argon generally available via Gemini API, AI Studio, and Gemini platform. The proprietary sparse mixture-of-experts frontier model supports native text, image, audio, and video input, with up to 1M tokens for combined input and output, including 1M-token responses. Introductory API pricing is $2/$10 per million input/output tokens, rising to $4/$20 after November 30, 2026. It leads on several benchmarks but trails rivals on some agentic coding and terminal benchmarks.
Liquid AI Releases LFM2.5-24B-A2B: A 2B-Active MoE for On-Device Agents
Liquid AI released LFM2.5-24B-A2B, an open-weight 24B-parameter MoE reasoning model with ~2B active parameters and 131,072-token context, targeting local agent loops. It runs via GGUF/MLX on laptops and mini PCs, with cited speeds of 184 tokens/s on Apple M5 Max and 114 tokens/s on AMD Ryzen AI Max+ 395, plus single NVIDIA B300 serving. It leads or matches larger MoEs on several agentic, tool-use, and instruction-following benchmarks, but trails Qwen3.6-35B-A3B on SWE-Bench Verified and GPQA Diamond. Weights are on Hugging Face under the LFM Open License v1.0, free for commercial use under $10M revenue.