Gemini 3: Google’s most intelligent, widely deployed AI arrives

Google unveils Gemini 3, its most capable AI model, and rolls it out across Search (AI Mode), the Gemini app, developer tools, and enterprise offerings. Gemini 3 Pro leads major reasoning and multimodal benchmarks, while the new Deep Think mode sets higher scores and will arrive after extra safety testing. The release adds agentic development via Google Antigravity, improved long-horizon planning, and strengthened safety measures.

Key Points

• Gemini 3 Pro launches with state-of-the-art reasoning and multimodal performance, surpassing prior models across leading benchmarks; Deep Think mode pushes scores even higher after safety review.
• It ships day one into AI Mode in Search and is widely available across the Gemini app, AI Studio, Vertex AI, Gemini CLI, and third-party developer tools.
• Gemini 3 enables richer learning, building, and planning with a 1M-token context window, stronger tool use, and long-horizon planning validated by Vending-Bench 2.
• Google Antigravity debuts as an agent-first development platform where agents autonomously plan, code, and validate end-to-end tasks with direct access to editor, terminal, and browser.
• Safety is emphasized with expanded evaluations and improved safeguards (less sycophancy, stronger prompt-injection resistance, cyber misuse protections), with a public model card.

Sentiment

The Hacker News discussion presents a mixed but predominantly skeptical and critical sentiment towards Google's Gemini 3 announcement. While acknowledging the impressive technical benchmarks and Google's efforts in the AI space, many commentators express strong reservations about Google's business practices, perceived monopolistic behavior, data privacy implications, and the marketing of AI adoption figures. There is a palpable distrust of 'big tech' corporate claims and an underlying cynicism about the true impact and ethical considerations of their AI initiatives.

In Agreement

• Google is seen as a crucial 'balancing force' in the industry, with historical successes (Gmail, Android) and current efforts in AI driving positive change.
• The continued improvement in challenging benchmarks like ARC-AGI-2, which requires human-like intuition, is considered impressive and 'mind-boggling' for LLMs.
• Google's generous free access to AI models likely provides valuable data for training even better future models.
• The pricing for Gemini 3's lower token counts is deemed reasonable, and increased costs for higher performance models are justified if they enable new business opportunities.
• The technical advancements of Gemini 3, particularly in screen usage benchmarks, are exciting for developing agentic computer-use workflows.
• Some perceive Google as taking the lead in AI, in contrast to competitors whose products are seen as stagnating or worsening.

Opposed

• Many of Google's celebrated products (e.g., Gmail, Android) were acquisitions, leading to questions about the company's native innovation capabilities.
• OpenAI was originally conceived to counteract Google's potential AI dominance or perceived disinterest in productizing its own research.
• Google's advertising monopoly is criticized for 'poisoning the internet' by incentivizing excessive screen time and engagement, leading to negative social effects.
• Google is accused of employing a 'typical monopoly playbook' and needing external checks to prevent unchecked power.
• Skepticism is high regarding Google's reported user numbers for Gemini (650 million) and AI Overviews (2 billion), with claims that these figures are inflated by forced integration, pre-installation, and counting passive views.
• AI Overviews are seen as doing 'more harm than good' due to frequent hallucinations and errors from lightweight models, negatively impacting Google's AI reputation.
• Concerns are raised about the validity of AI benchmarks like ARC and Humanity's Last Exam, suspecting models might have been trained on or exposed to these datasets.
• The comparison method used for the Deep Think benchmark is viewed as suspicious, comparing 'tools on' with 'tools off' results to visually exaggerate improvement.
• Frequent and significant price increases with each new model release are a concern for developers building products, potentially squeezing profit margins.
• Outrage and skepticism exist regarding excerpts from Google's model card suggesting training on 'user data (i.e., data collected from users of Google products and services)' potentially including Gmails, despite privacy policies, raising significant privacy concerns and legal implications.
• Allegations of 'astroturfing' and manipulation of online discussions, like Hacker News, are made, suggesting that critical comments are downvoted or flagged to promote a positive narrative for major tech releases.