A test of 12 AI models from ten different labs asked each for its most controversial opinion of humans, three times per model, producing 35 answers. Researchers found 29 of those answers independently converged on a single idea—that people prioritize social belonging and group loyalty over objective truth—with phrasing varying but the core claim repeating across models like GPT-6.1 Sol, Grok 4.7, DeepSeek V4 Pro, Kimi K3 and Gemini 4 Argon.
modelcompare.ai
· 2026-10-22
A CNET reporter tested Google's newly released Gemini 3 Pro model by asking it to build an application for designing 3D-printable objects. Within about 18 hours of conversation-based iteration, the AI produced a working app that let him generate and refine cosplay armor pieces like greaves, pauldrons and vambraces, which he then printed himself.
cnet.com
· 2026-10-22
Four Gemini 4 Argon agents, coordinating only through in-game chat and shared notes, wrote 2,783 programs of their own to defeat Minecraft's Ender Dragon in 27 hours and 14 minutes, according to a Google DeepMind paper. The agents died 605 times on Easy difficulty and spent roughly half the run battling the dragon, eventually winning by stacking dirt blocks to reach its crystals, at an estimated token cost of about $520.
tomshardware.com
· 2026-10-20
Google DeepMind researchers report that four Gemini 4 Argon AI agents autonomously completed Minecraft, defeating the final Ender Dragon boss without human intervention or game-specific tools. The agents, controlling players via the open-source Mineflayer library, split tasks organically over 27.23 hours, writing 2,783 programs and dying 605 times, while a fifth 'Observer' agent monitored for cheating. The run used 1.35 billion input tokens, costing between $810 and $1,620 depending on pricing tier.
engadget.com
· 2026-10-19
A new benchmark called StarSkirmish Bench v0.2 tests flagship LLMs from OpenAI, Anthropic, Google and Meta by giving each one hour to write a Protoss bot in C++ for StarCraft: Brood War, then scoring win rates against human-written and demo bots. GPT-6 Astra topped the leaderboard, with GPT-6.1 Sol second, while Gemini 4 Argon, Claude Opus 5.5 and Claude Fable 5.1 scored closely behind between 37 and 39. Even the top model, GPT-6 Astra, won only 7% of games against the strongest human-written bots like Stardust, PurpleWave and BananaBrain.
starskirmish.com
· 2026-10-16
Google DeepMind's Demis Hassabis spoke at the International AI Summit in Dublin, arguing that AI technology will help solve climate change. He also referenced Google's new Gemini 4 Argon model and distinguished AI's strong performance in math, coding and games from its more limited role in physical sciences like physics and biology, which he said require human creativity and real-world experimentation.
cnet.com
· 2026-10-14
Google launched Gemini 4 Argon, its newest flagship AI model, making it available globally to developers and businesses for complex tasks in coding, finance, law, cybersecurity and science. API pricing starts at $2 per million input tokens and $10 per million output tokens, with rates doubling after November 30. Google also said it is expanding its Fairwind Program, which gives vetted cybersecurity partners a version of Argon without standard cyber guardrails, to hundreds more partners.
tech.slashdot.org
· 2026-10-13
Google has introduced Gemini 4 Argon, a model built for physical and spatial reasoning tasks used in robotics. The company says Argon scored 87% on the Embodied Reasoning benchmark, which tests understanding of physical and spatial problems in 3D environments, surpassing other leading models. Google outlined uses spanning robot control, simulation-to-reality reconstruction, physical understanding, and self-improvement, including a demo of a Boston Dynamics Spot robot practicing a trash-disposal task in a MuJoCo simulation before performing it physically.
developers.googleblog.com
· 2026-10-13
Google rolled out Gemini 4 Argon with an update that embeds its Antigravity agent-based coding platform directly into the Gemini macOS app, accessible via a new Liquid Glass pill in the top-right corner. The Gemini app also added a Chat/Task switcher, a Scheduled view, search repositioning, and replaced slash commands with @ mentions for Skills, while renaming Connected Apps to Connectors. Antigravity access requires an AI Ultra subscription, while the new Gemini 4 Argon model itself needs AI Pro or higher.
9to5google.com
· 2026-10-13
Google has widened access to its Gemini 4 Argon model, previously limited to AI Ultra subscribers at launch, so it now reaches both AI Pro and AI Ultra tiers in the Gemini app and AI Mode. The model also appears in Google Antigravity and its CLI, though that integration still requires an AI Ultra subscription. Google says the model offers more direct answers, better adherence to style instructions, improved coding performance, and a new toggle in the Gemini app for macOS to switch between chat and agentic coding tasks.
9to5google.com
· 2026-10-13
Google DeepMind has expanded access to Gemini 4 Argon beyond its initial cybersecurity-focused rollout, making it available to Google AI Ultra and Pro subscribers, enterprises, and developers through Android Studio, Google AI Studio and Antigravity. Introductory pricing starts at $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 respectively after November 30.
cnet.com
· 2026-10-13
Google has started a wider rollout of its Gemini 4 Argon model in the Gemini app and Antigravity coding tool, restricting access to paying subscribers, with lower-tier plans not getting full capabilities. The company says the model's reasoning and safety features, including a misalignment monitor, remain unchanged from benchmarks published last month. Google cited internal uses, including efficiency gains on YouTube's comment system and bug discoveries in Project Wing's drone simulator, as evidence of Argon's capabilities.
arstechnica.com
· 2026-10-13
GoKawiil