nanomangonanomango

Feed your brain. Skip the noise.

AI today in 60 seconds

Today's brief — full text

All five stories from the home strip, with full articles. Scroll to read each one.

← Back to home

Gemini 3.6 Flash Cuts Price, Teases Gemini 4

AI-assisted

Google's new workhorse model uses up to 17% fewer tokens and hints at Gemini 4.

According to TechCrunch and 9to5Google, Google released Gemini 3.6 Flash on July 21 alongside 3.5 Flash-Lite and a cybersecurity-tuned 3.5 Flash Cyber. Google positions 3.6 Flash as its everyday workhorse, with better coding, knowledge work and multimodal performance while cutting token use by up to 17%, which makes it cheaper to run than the model it replaces. It is available now in the Gemini app, the API via Google AI Studio, and the enterprise agent platform. Notably absent was the much-delayed Gemini 3.5 Pro; instead Google teased a larger Gemini 4 release. For anyone building on Gemini, the takeaway is a cheaper default tier and a signal that the next flagship is close.

Sources

Apple Passes Nvidia as Chip Stocks Enter Bear Market

AI-assisted

The SMH ETF fell 8.9%, its worst week since April 2025, as AI capex fears bite.

CNBC reports that chip stocks just had their worst week in over a year, with the VanEck Semiconductor ETF sliding 8.9% — its steepest weekly drop since April 2025. The Philadelphia Semiconductor Index fell nearly 10% and is now down more than 20% from its late-June high, confirming a bear market. Nvidia shed 2.2% on Friday, and Apple overtook it as the most valuable US company. The trigger, per CNBC, is investor unease about whether the roughly $725 billion being poured into AI infrastructure will pay off. Analysts quoted argue fundamentals look intact and see this more as a valuation reset than a collapse in AI demand. Worth watching: whether Big Tech earnings this week calm or deepen the nerves.

Sources

No AI Lab Beats a C+ on Safety, Anthropic Leads

AI-assisted

Anthropic scored 2.66/4; xAI, DeepSeek and Mistral got outright failing grades.

The Future of Life Institute published its Summer 2026 AI Safety Index, and as TechTimes notes, no lab scored above a C+. The index graded nine leading developers across 37 indicators in six domains. Anthropic topped the field at C+ (2.66 on a 4.0 scale), with OpenAI and Google DeepMind tied at C, while xAI, DeepSeek and Mistral received failing grades. The report's sharpest point is a trend: amid fierce competition and a push into defense contracts, several labs have quietly walked back earlier commitments to pause development when systems approach danger thresholds. For anyone weighing which AI tools to trust, it is a reminder that even the safest-rated lab is only a middling C+.

Sources

EU Orders Google to Open Android to Rival AI Apps

AI-assisted

Gemini's system-level Android access must open to ChatGPT and Claude by 2027.

As TechTimes and Unite.AI report, the European Commission adopted two binding Digital Markets Act decisions ordering Google to give rival AI assistants and search engines the same access it reserves for its own products. Today Gemini works as a built-in Android feature that responds to a wake word, sees the screen, and acts across apps; the ruling opens those system-level hooks to competitors like ChatGPT and Claude. Google must also share anonymized search query, click and ranking signals on fair, non-discriminatory terms. Data sharing is due to begin by January 2027, with broader Android interoperability reaching users from July 2027. Fall short, and the Commission can pursue penalties of up to 10% of global revenue.

Sources

Salesforce Backs 8090's $135M AI Coding Bet

AI-assisted

The startup's platform lets teams build enterprise software alongside AI agents.

According to this week's startup funding roundups, 8090 Solutions raised a $135M Series A backed by Salesforce, Craft Ventures and the Launch fund. The company builds a platform for developing enterprise software in collaboration with AI agents, part of a wave of tooling that aims to move beyond code autocomplete toward agents that take on larger chunks of a build. The round stands out for its size at Series A and for Salesforce's involvement, signaling that a major enterprise-software incumbent sees value in agent-driven development. For engineering teams, it is another sign that 'agents that build software with you' is becoming a real product category rather than a demo.

Sources

All news →