Post
14 Oct 2026
Liquid AI releases LFM2.5-24B-A2B for on-device agents
Open-weight 24B MoE with ~2B active params and 131K context. Liquid reports LiveClaw 48.09 vs Qwen3.6-35B's 38.19, AutomationBench 27.69 vs 8.72, 184 tok/s on Apple M5 Max, and 21K+ output tok/s on one NVIDIA B300.
02RoboticsGemini 4 Argon taught a humanoid cornhole from one video
In an internal test Argon turned a raw human video into simulation code; Apptronik's Apollo then sank 7 straight shots in a real-world trial. Google says Video-to-Sim-to-Real compresses weeks of simulation modeling into hours.
Google · blog.google ↗ (opens in a new tab)
03SecurityGitHub swaps an LLM for a small ModernBERT model in secret scanning
Regex candidates + fine-tuned ModernBERT classifier: 11.5% higher owner-weighted acceptance, 86% lower median scan time and 98%+ lower serving cost vs the previous LLM pipeline; offline 13% higher precision, 19% higher recall. Now extends generic detection to push protection.
GitHub Blog · github.blog ↗ (opens in a new tab)
04VisionPrime Intellect acquires Moondream; 3.1 and Photon now live
Founders Vik Korrapati and Jay Allen join to lead inference and multimodal work. Moondream 3.1 is live on Prime Inference via Photon; Prime cites 34 req/s (~24x the next most accurate open model) and 60+ inferences/sec on one H100. Moondream stays the brand for small multimodal models.
Prime Intellect · primeintellect.ai ↗ (opens in a new tab)- 05Vision05Vision
Datalab Indra OCR: 1M params, 49 pages/sec on CPU
Indra OCR reads 49 pages/sec on 32 CPU cores with no GPU (11–17x vs Tesseract and PaddleOCR PP-OCR text models), outputs line boxes, layout labels and reading order as markdown/JSON, and beats PP-StructureV3 on OmniParseBench headline (62.1 vs 58.4) at ~1/180 the time per page. Code Apache 2.0; weights modified OpenRAIL-M.
Datalab · datalab.to ↗ (opens in a new tab) - 06Open Models
Google DeepMind's fine-tuned Gemma spots scams on-device
Built with Singapore's GovTech: Gemma fine-tuned on local scam conversations flags manipulation on the phone, so encrypted chats aren't scanned in the cloud. A lightweight classifier screens messages, then Gemma checks the wider conversation. Training excluded private/PII data.
Google DeepMind · deepmind.google ↗ (opens in a new tab) - 07Models07Models
OpenAI retires GPT-4o and GPT-4.1 from its API on Jan 14, 2027
Deprecated Oct 14: gpt-4o (2024-08-06, 2024-11-20), gpt-4o-mini, gpt-4.1 and gpt-4.1-mini shut down January 14, 2027; fine-tunes follow the same schedule. Recommended replacement for all: gpt-6-luna.
OpenAI · developers.openai.com ↗ (opens in a new tab) - 08Research08Research
Modal trained RL across 6 GPU types — and it stayed stable
Mixed-precision, mixed-accelerator samplers (B200/B300 NVFP4, H100/H200 FP8+BF16, A100/RTX PRO BF16) with off-policy correction kept GRPO stable; SWE-bench Pro V2 on par with homogeneous B200 despite 4.7x larger KL; ~15% lower sampler cost per update.
Modal · modal.com ↗ (opens in a new tab) - 09Tools09Tools
Google Play installs Android apps you vibe-code with AI
AI Studio vibe-coded Android apps can launch privately on your device via Play with a standard Google account — no developer account; rolling out in select markets including the U.S., 18+.
Google · blog.google ↗ (opens in a new tab)