Cognition's Devin Cloud Lets Developers SSH Into AI Coding Sessions
Cognition wired Devin Cloud into the CLI with session handoffs and full SSH access to the agent's VM, blurring local and remote work.
// Intelligence
Today in AI
A focused feed of models, agents, research and open-source releases for people building with AI.
Updated
Liquid AI's LFM2.5 Tops Mobile AI Charts Using Half the Memory
Liquid AI's LFM2.5 family posts Pareto-frontier scores on Artificial Analysis's new mobile benchmark, matching a top 3B model at 40% less memory.
Kyutai's Voice of Reason Thinks While It Speaks, Hitting 77% Math Accuracy
Kyutai's Voice of Reason turns GLM-4-Voice into a speech-native math solver, jumping GSM8K accuracy from 27.3% to 77.1% with no added latency.
SGLang Squeezes 78% More AI Throughput on NVIDIA's Blackwell GPUs
SGLang, Qwen, and NVIDIA ship 4-bit NVFP4 KV cache on Blackwell, packing 1.78x more context and boosting long-context decode up to 78%.
Krea Opens 60-Plus AI Image and Video Models to Coding Agents
Krea's hosted MCP server lets Claude, Cursor, Codex, and other agents generate images, video, and upscales from plain prompts.
Perplexity Computer Adds MiniMax H3 and Seedance to Build Videos in One Thread
Perplexity Computer now generates finished video clips inline via MiniMax H3 and ByteDance Seedance 2.5, bundled with copy and creative for Pro and Max users.
xAI Ships Grok 4.7 With Longer Agent Runs at Unchanged Prices
xAI ships Grok 4.7 with longer deliberation, better self-checking, and stronger safety guardrails, matching Grok 4.6 on price and latency.
Understanding the “Harness Tax” Behind Coding Agents
21 combos, 60 tasks, 5x cost swing, and a 4-tool harness on the Pareto frontier
Microsoft's RetroChimera Plans Drug Synthesis Routes and Wins 9 of 10 Expert Reviews
Microsoft Research's RetroChimera combines two neural networks to plan chemical syntheses that expert chemists prefer over published literature reactions.
Boston Dynamics Opens Atlas Training Facility Inside Hyundai's Georgia Factory
Boston Dynamics opens a dedicated Atlas training center inside Hyundai's Georgia EV plant, targeting 25,000 humanoids deployed across factories.
Moonshot AI Ships Kimi Code Desktop With 1M-Token AI Coding Agent
Moonshot AI ships a desktop coding agent for macOS and Windows, wrapping the Kimi Code CLI in a workspace with parallel agents and a built-in browser.
Unitree's Dex5-S Brings a 22-DoF Robotic Hand to Developers for $6,500
Unitree's new Dex5-S packs 22 backdrivable joints into a human-sized robotic hand starting at $6,500, targeting the dexterity bottleneck holding back humanoids.
Sac-Y's Open-Source Jev-cu Makes Computer Agents 200x Faster and Cheaper
A new open-source project routes computer-use decisions through TypeSafe's Jev model, letting a text-only classifier pick UI actions while Codex executes them.
Jared Palmer's Open-Source Kev Answers Typed Questions in one Pass
Jared Palmer's open-source Kev family scales a Jev-style decision model up to 8B parameters using Qwen3, LoRA, and a pointer head.
Z.ai Opens ZCode's Full Coding Agent Stack to Developers Worldwide
Z.ai just open-sourced ZCode, a full coding agent workbench spanning desktop, web, and terminal, tuned for GLM-5.3 but tool-agnostic.
Bespoke Labs' Nimble Beats a 27B Model at Classification Using Just 9B
Bespoke Labs open-sources Nimble, a 9B model that makes typed decisions without generating tokens, using contrastive pairs for training.
OpenHuman Packs Local AI Memory and Agent Graphs Into one Rust App
OpenHuman is an open source, local-first agent harness that gives your AI a persistent memory of your life and orchestrates fleets of agents on your machine.
Open-Source Laya Beats Closed Jev API at 7x Faster Decision Speed
Laya is an open-source, non-autoregressive decision engine that answers typed questions across 100+ languages in a single 33ms forward pass.
Alibaba's Qwen-Image-2.1 Ships Transparent Image Generation Without Extra Tools
Alibaba open sources a 7B image model with native RGBA output, 10 reference image editing, and unified generation plus editing capabilities.
Andrew Levin's Open-Source macOS Agent Makes Decisions 155x Cheaper Than Claude Opus
A macOS agent skips screenshots to frontier models, uses OCR plus a tiny classifier called TypeSafe, and runs steps for a fiftieth of a cent.
Laya Runs AI Decisions in 7ms on Apple Silicon Without Cloud APIs
A native MLX port of the Laya typed decision model runs 7-14 ms structured classifications locally on Apple Silicon, no text generation required.
HyperQwen Runs Qwen3.8-27B on a Single RTX 3090 at 1,035 tok/s
HyperQwen crams a 27B Qwen model onto one 24GB RTX 3090 with vLLM patches, hitting 127 tok/s solo and ~1,035 tok/s at 64 concurrent.
Tencent's AuK Handles 16 Audio Tasks From a Single Prompt
Tencent Hunyuan open-sources a 1.5B speech foundation model that handles TTS, editing, enhancement, and separation through plain-language instructions.
Alibaba's Qwen3.8 LiveTranslate Cuts Speech Translation Lag to 2.3 Seconds
Alibaba's next-gen simultaneous interpretation model cuts average lag to 2.3 seconds across 60 languages, adds speaker diarization, and clones each voice separately.
No stories match these filters.