You can now choose how hard Claude thinks before answering your queries
Anthropic's Claude Opus 4.8 update lets you control how hard Claude thinks before it answers.
AbJimroe
For the first time, Claude users can decide whether their AI assistant thinks fast or thinks deep.
Shikhar Mehrotra / Digital Trends
Anthropic just released Claude Opus 4.8, and while the benchmark improvements are quite real, the most meaningful change for everyday users is something far simpler.
You can now tell Claude how hard to think before it responds to your query. Along with that, dynamic workflows are now available in research preview for Enterprise, Team, and Max plan users.
Fast mode is available for Opus 4.8. It’s the same model at roughly 2.5x the speed, and we’ve made it three times cheaper than before.
Turn it on with /fast in Claude Code. On the API, contact your account manager to request access or join the waitlist: https://t.co/5SLFF77Whn
What is effort control and why does it matter?
A new control sitting alongside the model selector on claude.ai page, where you interact with the AI model. It lets you choose the amount of effort Claude applies to any given query.
There are five effort settings to choose from when using Opus 4.8: Low, Medium, High (Default), Extra, and Max. As the name suggests, low effort responses are faster, and therefore, better for quick or trivial questions, email drafts, or anything else that requires speed over depth.
The High and Max settings, on the other hand, force Claude to think more deeply before responding. These effort levels are ideal for complex multi-step problems, detailed analysis or comparison, and anything where accuracy matters more than speed. However, as the AI model takes more time to respond, these effort levels exhaust your rate limits faster.
The control is available on all plans. Until now, you only submitted a prompt, selected the AI model based on its known strengths, and Claude decided how much thinking to apply. Effort control effectively hands that decision to you.
Claude
What else changed?
The most significant practical upgrade for developers is dynamic workflows. Claude Code can now plan a large task, spin up hundreds of parallel subagents in a single session, and verify its own outputs before reporting back.
The Messages API now accepts system entries inside the messages array, allowing developers to update Claude’s instructions mid-task without breaking the prompt cache. This is especially useful for updating permissions, token budgets, or environment context.
Anthropic also mentions that Opus 4.8 is around four times less likely than its predecessor to let flaws in code pass without noticing. The release arrives 41 days after Opus 4.7, reflecting an accelerating update frequency.
For more than five years, Shikhar has consistently simplified developments in the field of consumer tech and presented them…
Copilot gets a redesign and it now wants to do more without being an eyesore
Microsoft is giving Copilot a quiet but meaningful redesign, and this time the focus is not just on making it more powerful. It is about making it feel like something that naturally belongs in your workflow.
Across Microsoft 365, Copilot is being reshaped to reduce visual noise and increase usefulness. Instead of constantly demanding attention, it is being designed to sit in the background when needed and step forward only when it actually helps. That shift might sound subtle, but in day-to-day work, it changes how often you feel interrupted versus supported.
Just like humans, this robot can hear music and play it after just two minutes of self-practice
The successful experimentation points toward a new model for rehabilitation robotics, based on experiential learning.
In a neuro-robotics lab at the University of Southern California, a small mechanical hand heard a melody for the first time and played it back in a single attempt, without any sheet music, pre-loaded scores, or weeks of supervised training and practice (via USC Viterbi).
The system is called the Musician Hand. It has four fingers, each moved by a tendon connected to a small electric motor, mirroring how muscles actually pull tendons in a human hand. It was built by doctoral candidate Hesam Azadjou under the direction of Professor Francisco Valero-Cuevas.
As AI spills everywhere with quick answers, research finds that the internet’s soul is dying
AI gives you answers fast, but a new study says something important gets lost along the way.
A new collaborative study by UC Riverside computer and social scientists has found that as people increasingly rely on AI for answers, the internet risks losing the very thing that made it interesting in the first place: human emotion, lived experience, and messy, opinionated thinking.
The study compared how AI systems like ChatGPT and Gemini respond to subjective questions versus traditional web searches. The researchers asked both AI and web search engines opinion-heavy questions, such as whether governments should ban fossil-fuel cars or whether the US healthcare system needs reform, and analyzed the reasoning behind each response.
/cdn.vox-cdn.com/uploads/chorus_asset/file/22481672/acastro_210430_1777_semiCon_0001.jpg)
