AI
Artificial intelligence: LLMs, agents, diffusion, vision, NLP and the latest from top labs.
164 published articles
AI Agents
Microsoft's bet on small models for agentic AI is about orchestration, not knowledge
Microsoft Research's MagenticLite project shows that small models can handle complex agentic tasks when orchestration, model design, and execution environment are codesigned. The release includes MagenticBrain, a 14B orchestrator, and Fara1.5, a computer-use model family that nearly doubles prior performance on web navigation benchmarks.
2026-07-09
Artificial Intelligence
Nvidia's new audio model does five jobs at once and beats the specialists at their own game
Nvidia's Audex unifies audio understanding, generation, and text reasoning in a single model, matching or beating task-specific systems on speech and audio benchmarks without sacrificing text performance.
2026-07-09
Edge AI
Free GPU, no cloud: this developer just made Hindi LLMs work on a laptop
A developer successfully fine-tuned Google's Gemma 4 for Hindi on a free GPU and shipped the model to run on a CPU. The case study demonstrates practical techniques for low-resource language adaptation and edge deployment.
2026-07-09
API economics
DeepSeek V4 drops in July with a pricing shift that will hurt if you code when the sun is up
DeepSeek V4 arrives mid-July with peak-load pricing. The API costs double during business hours, up to $1.74 per million output tokens, but off-peak rates can be a steal for cache-hit workloads.
2026-07-09
Alibaba Cloud
Alibaba Cloud's EMR Serverless Spark now processes images and video in plain SQL, no Python needed
Alibaba Cloud’s EMR Serverless Spark now supports images and video frames directly in SQL, letting data engineers skip Python overhead. A case study on autonomous driving data preprocessing shows automated ETL pipelines powered by Qwen vision models replacing manual annotation.
2026-07-09
Coding Agents
Cognition's new coding agent scores near frontier results for pocket change
Cognition's SWE-1.7 coding model narrows the gap to frontier systems at a fraction of the cost, scoring 42.3% on FrontierCode and running at 1,000 tok/s. The model was trained on an improved reinforcement learning pipeline using Kimi K2.7 as the base.
2026-07-09
Software development
The hidden tax on vibe-coded projects that shows up when you least expect it
An in-depth analysis of the promise and peril of vibe coding: why AI-generated code accelerates prototyping but introduces security risks, performance limits, and technical debt that demand careful human oversight.
2026-07-09
AI Tools
Kimi slides takes on every file format you throw at it, and that is the easy part
Kimi Slides covers the full workflow from research and content structuring to layout and formatting, accepting text, documents, images, or your own PowerPoint template. Complex diagrams are rendered as editable native components, and on-slide citations make the content verifiable.
2026-07-09
AI Agents
Parallel agents aren't about speed. They're about coordination.
Parallel agents split complex tasks across concurrent workers, each with isolated state and defined scope. The real challenge isn't speed, it's coordination.
2026-07-09
Artificial Intelligence
Kimi K2.7 Code is faster and cheaper. But open-source coding just hit a wall called GPT-5.5.
Moonshot AI's Kimi K2.7 Code makes big gains on long-horizon coding tasks with 30% less token waste. Yet GPT-5.5 and Claude Opus 4.8 still lead on key benchmarks, highlighting the real-world trade-offs of open-source decisions.
2026-07-09
Open-source AI
China's MiniMax just open-sourced a 1M-token model that outruns GPT-5.5 on real coding tasks
MiniMax M3 is the first Chinese open-source model to combine native multimodality, 1M-token context, and advanced agentic coding. Its structured approach to scale, including a new attention architecture called MSA, challenges the assumption that open models must trail behind proprietary systems.
2026-07-09
AI Labs & Research
MiniMax's M3 just wrote its own CUDA kernel, and opened the code
MiniMax M3 scores 83.5 on BrowseComp, edges past Opus 4.7, and handles up to 1M tokens natively. In a remarkable autonomy test, it self-optimized a GPU kernel from 7.6% to 71.3% peak utilization without human intervention.
2026-07-09