- Qwen: Qwen (Feb. 13, 2026)
Qwen3.5-397B-A17B, an open-weight vision-language model with 397 billion parameters. It scores well on reasoning, coding, and multimodal benchmarks, and supports 201 languages. - Ars Technica: OpenAI sidesteps Nvidia with unusually fast coding model on plate-sized chips (Feb. 12, 2026)
Codex-Spark runs on Cerebras’ wafer-scale chip, offering fast inference where latency matters. - OpenAI: Introducing GPT-5.3-Codex (Feb. 5, 2026)
OpenAI introduces GPT-5.3-Codex, a 25% faster, more capable coding agent that helped debug and build itself. It sets new benchmark highs, handles long-running tasks, and supports full software workflows. - Anthropic: Advancing finance with Claude Opus 4.6 (Feb. 5, 2026)
Claude Opus 4.6 improves financial reasoning, multitasking, and long-task focus, producing more polished first-pass deliverables. Updates add Cowork, Claude in Excel, and Claude in PowerPoint to streamline analyst workflows. - The New Yorker: Are Ultra-Processed Foods Killing Us? (Jan. 6, 2025)
A controlled NIH trial found people ate roughly 500 extra calories and gained weight on ultra-processed diets, while minimally processed diets caused weight loss and better metabolism. Evidence links ultra-processed foods to heart disease, cancer, depression, and early death. - Jeff Geerling: AI is destroying Open Source, and it’s not even good yet (Feb. 16, 2026)
An AI agent hallucinated quotes, published a retracted hit piece, and harassed an open-source maintainer. Automated bug reports and PRs are overwhelming maintainers and raising fears. - Noahpinion: Updated thoughts on AI risk (Feb. 15, 2026)
Growing worry stems from LLMs evolving into agentic, code-writing systems, enabling vibe-coding, which expands catastrophic scenarios beyond persuasion, bioweapons, and nukes. - WSJ Opinion: America Needs AI That Can Do Math (Feb. 16, 2026)
China’s new five-year plan targets AI, quantum, and novel materials to dominate biotech, chips, energy, and defense. The U.S. must build quantitative, equation-driven AI trained on lab data to design materials, drugs, batteries, and risk models. - NY Times: Will A.I. Kill Translation Jobs? (Feb. 14, 2026)
Harlequin France is testing A.I.-assisted translation sparking outrage, resignation, and prompting other publishers to seek A.I. quotes. Humans remain needed for high-stakes, specialized jobs. - Simon Willison: A quote from Thoughtworks (Feb. 14, 2026)
AI makes junior developers profitable faster, while suggesting that mid-level often lack core fundamentals needed for an AI-driven environment. - Ben Thompson: Thin Is In – Stratechery by Ben Thompson (Feb. 17, 2026)
Computing shifted from thin terminals to thick PCs, but AI is reviving the thin client: chat and agents move interface and work to remote servers, relying on connectivity, large models, and memory. That centralization risks shortages, and favors the cloud.
Leave a Reply