OpenAI’s model picker complicates ChatGPT, Elon Musk targets Apple with a lawsuit, and Anthropic acquires Humanloop’s team in the competitive AI talent race.
OpenAI’s GPT-5 rollout aimed to simplify ChatGPT with an auto model router but instead reintroduced complexity, adding “Auto,” “Fast,” and “Thinking” modes plus legacy models like GPT-4o. User backlash and attachment to older AI personalities highlight demand for customization.
Perplexity AI has made a surprise $34.5B bid for Google Chrome, far below estimates of its true value. Critics call the move a stunt given Chrome’s 3B users and Perplexity’s $18B valuation, though the startup claims it seeks to promote open web access and user safety.
Elon Musk threatened legal action against Apple, accusing it of unfairly promoting rival AI apps over his Grok chatbot and X, calling it an antitrust violation. Apple denied bias, while context highlights Musk’s past disputes with Apple, OpenAI, and manipulation claims at X.
xAI co-founder Igor Babuschkin has left the company to launch Babuschkin Ventures, a firm backing AI safety and startups, after helping build xAI into a leading model developer. His exit follows Grok scandals, though xAI remains competitive against OpenAI, DeepMind, and Anthropic.
Anthropic has acqui-hired Humanloop’s co-founders and team to strengthen enterprise AI tools, focusing on evaluation, safety, and reliability. The move bolsters Anthropic’s enterprise growth and safety-first strategy as it competes with OpenAI and DeepMind.
Anthropic added a feature allowing Claude Opus 4 and 4.1 to end chats in rare, extreme cases of harmful interactions, citing precautionary concerns over “model welfare.” The capability is experimental, limited to last-resort use, and does not apply in self-harm scenarios.
Meta has lost much of its top AI talent to rivals, with insiders describing chaotic culture, failed projects like Llama 4, and reliance on huge pay offers to lure researchers. Despite denials, critics say Meta overpays for mediocre hires while rivals attract “missionaries” driven by purpose.
Anthropic’s Claude Sonnet 4 now supports a 1M-token context window, enabling full codebase analysis, large-scale document synthesis, and advanced agents. Pricing scales with longer prompts, with caching to reduce costs. The update is in beta on Anthropic API and Amazon Bedrock, with Google Cloud coming soon.
Gemma 3 270M is a compact, energy-efficient model optimized for fine-tuning and instruction following, ideal for specialized tasks like text processing, compliance, and creative apps. It enables fast, low-cost on-device deployment, expanding the Gemma ecosystem for task-specific AI.
DINOv3 is a self-supervised vision model that outperforms supervised and CLIP-based approaches in tasks like segmentation, detection, and depth estimation. Trained on 1.7B images, it scales to 7B parameters, offers smaller efficient variants, and is already aiding WRI climate monitoring and NASA robotics.
A feature-rich gateway, proxy and MCP Registry that federates MCP and REST services
LLM agents built for control, designed for real-world use, and can be deployed in minutes.
Claude Code now supports a new model option "Opus Plan Mode" that will use Opus 4.1 for planning and Sonnet 4 for execution.
Claude Code now lets you customize communication styles with /output-style and lets you pick between Explanatory and Learning output styles.
Based on feedback, GPT-5 felt too formal. The implemented changes are sublte, but GPT-5 should feel more approachable now.
Cursor CLI now includes MCPs, Review Mode, /compress, @-files, and other UX improvements.
GPT-5 improves agentic workflows, coding, intelligence, and steerability, with new features like minimal reasoning, verbosity control, and tool preambles. Prompting strategies help balance autonomy and efficiency, while case studies like Cursor show how tuning boosts coding clarity and task completion.
The future of software engineering is unfolding with the rise of AI teammates: autonomous, goal-driven systems collaborating with human developers. Among these, autonomous coding agents are especially transformative, now actively initiating, reviewing, and evolving code at scale. This paper introduces AIDev, the first large-scale dataset capturing how such agents operate in the wild.
A curated weekly roundup of AI news, tools, and breakthroughs to stay ahead of the curve in a fast-moving industry.
Curated by devs, for devs. Helping 400+ developers stay ahead of the AI curve.