Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Gemini 4 Argon

AI overview

Sign in and the AI will write an overview from our coverage.

Headlines · 121

  1. Four Gemini agents reportedly defeat Minecraft’s Ender Dragon

    Google DeepMind reportedly deployed four Gemini 4 Argon agents in Minecraft, where they began without human-written gameplay code and learned to gather resources, craft tools, coordinate roles, and recover from failures. The reports say the agents wrote more than 2,700 programs and defeated the Ender Dragon after 27 hours of real time, following more than 80 days of experimentation. The details come through media summaries of a DeepMind paper rather than an independently reviewed paper in the supplied reports.

    虎嗅 AI · 🔥 0
  2. Gemini Android app updates side panel and settings

    The Gemini app has rolled out redesigned side-panel and settings pages for Android following a server-side update, according to 9to5Google. Search has moved to the top-right corner, while the navigation drawer shows five items before expanding through a “More” menu. Google has consolidated preferences into one settings page that includes a new Theme option. The updated design had previously rolled out to iOS, macOS and the web alongside the launch of Gemini 4 Argon.

    9to5Google · 🔥 0
  3. Epoch AI Finds Models Cannot Yet Reliably Automate Its Knowledge Work

    Epoch AI researchers Kelly Hong and Greg Burnham gave six leading models 11 realistic tasks drawn from work at the organization, covering graphic design, short research reports, interactive data tools, AI data-center research, and research design. Claude Fable 5.1 and GPT-6 Astra were roughly tied for the lead, but no model reliably met Epoch’s quality standards; closed-weight models performed best on well-defined tasks, while open-ended work requiring broader judgment remained difficult.

    Epoch AI · Gradient Updates · 🔥 0
  4. Google Releases Gemini 4 Argon for Complex Coding and 3D Projects

    Google released Gemini 4 Argon, a frontier model intended for multi-step work across coding, finance, legal, cybersecurity, and science. The model is globally available to developers and enterprises through the API in AI Studio and the Gemini platform, while Google AI Ultra and paid Pro subscribers can access it in Antigravity, the Gemini app, and AI Mode in Search. Google also showcased five community-built projects, including a 3D earthquake globe, a jet-engine simulation, and a multiplayer browser game.

    Google Gemini Blog · 🔥 0
  5. Gemini 4 Argon Builds a 3D Gothic Cathedral in Antigravity

    Gemini 4 Argon built a 3D Gothic cathedral in Antigravity using a single Nano Banana image and Blender MCP, according to the report. The model iterated on geometry, textures, and lighting to create the finished scene.

    Google DeepMind (X) · 🔥 0
  6. Frontier AI Models Used in Reported PHP RCE Exploit Development

    A report says security teams from Calif and Wiz used frontier AI models to find two PHP remote-code-execution vulnerabilities and build exploits. The flaws, CVE-2026-109618 and CVE-2026-109619, are marketed as PHP2Shell. The report says a 2024 PHP Foundation audit had identified CVE-2026-109618 but could not determine how to create an exploit from the theoretical issue or agree on a fix. It cites an estimate that PHP2Shell could affect as much as 70% of the internet, including thousands of major websites, though the report does not substantiate that estimate. There is also disagreement over whether working exploit attempts occurred before the patch was released.

    Hacker News 首页(100+ 分) · 🔥 0
  7. AI Models Helped Discover PHP2Shell Vulnerabilities

    Wiz researchers said they used Wiz Atlas, Gemini 4 Argon and GPT-5.6 Sol to identify two PHP vulnerabilities in late August, then validated them in Drupal and WordPress before reporting them to the PHP development team. The flaws date back at least to PHP 5.4, released in 2012. PHP subsequently released patches for the two vulnerabilities, both rated critical under CVSS v4.0. Accounts differ on whether exploitation occurred before disclosure: Amazon said it first detected activity on October 17, while CISA said it had not observed exploitation.

    iThome 台湾 · 🔥 0
  8. Gemini Personal Intelligence Adds Google Maps Context

    Google is expanding Personal Intelligence in the Gemini app to use Google Maps searches and saved places alongside Google Calendar, according to Google Gemini and 9to5Google. With user permission, the feature can combine a user’s schedule with saved locations to recommend nearby restaurants or other options. Separately, Google is notifying AI Ultra subscribers that Gemini 4 Argon will replace Deep Think soon, but the notice does not clarify whether Gemini 3.1 Pro is being retired or only the reasoning mode is being phased out.

    9to5Google · 🔥 0
  9. 12 AI Models Share a Conformity-Themed View of Humans

    R1’s test team asked 12 text models from 10 laboratories in three countries for their most controversial, unhedged take on humans three times. Every model gave a conformity-themed answer at least once, saying people prioritize group beliefs, belonging, status or personal stories over truth or principles; seven of the 12 gave that shared theme in all three answers. The study used no system prompt or retries and made 60 calls costing $1.54.

    Hacker News · AI(100+ 分) · 🔥 0
  10. Gemini 4 Argon Builds a 3D-Printable Armor App

    CNET managing editor James Bricknell tested Gemini 4 Argon by asking it to build a web app for modeling and 3D-printing armor. According to his account, the model created the KeriForge app through about 10 iterations, and he went from initial screenshots to a physically printed object in under 18 hours.

    CNET · 🔥 0
  11. Gemini 4 Carbon reportedly enters partner early access and LM Arena testing

    Reports say Gemini 4 Carbon has entered an invite-only partner early-access program and is being tested on LM Arena, but it has not been formally released. Testers described Carbon as noticeably better than Gemini 4 Argon at 3D generation, including a complete Super Mario level, while reporting that it was roughly two to four times slower depending on the task. They also described its SVG and frontend/UI output as weak. The comparisons come from arena matchups and hand-picked prompts rather than standardized benchmarks, and the checkpoint served on LM Arena cannot be independently verified.

    DoNews · 🔥 0
  12. GPT-5.6 Sol helped find two critical PHP flaws, now patched

    The Rundown AI reported that OpenAI's GPT-5.6 Sol helped researchers identify two critical flaws in PHP, the language behind roughly 70% of websites, and that PHP maintainers have now released patches. The newsletter said security firm Calif found the first bug on Aug. 1 using GPT-5.6 Sol, while Google-owned Wiz separately caught both flaws weeks later with Sol and Gemini 4 Argon. According to the report, OpenAI said it worked with PHP maintainers on patches for two months and called the find one of the most impactful security issues in recent memory; the NSA urged organizations unable to patch to take systems offline, and Cloudflare said it was seeing early attacks.

    The Rundown AI · 🔥 0
  13. AI-found PHP2Shell flaws under mass exploitation as PHP ships emergency patches

    PHP released versions 8.2.35, 8.3.36, 8.4.27 and 8.5.12 on October 21 to fix two memory flaws collectively called PHP2Shell, CVE-2026-109618 and CVE-2026-109619, both rated 9.2 under CVSS v4.0. The PHP Foundation said security firm Calif used OpenAI's GPT-5.6 Sol on August 11 to confirm the bugs could yield unauthenticated remote code execution, while Wiz independently found the first flaw using Gemini 4 Argon and GPT-5.6 Sol. Hadrian reported mass exploitation attempts, mainly through PHP's session upload-progress mechanism, about two and a half hours before the patches appeared, and Amazon said attackers tried to exploit the flaws four days before public disclosure.

    iThome 台湾 · 🔥 0
  14. Google Cloud details governed self-evolving AI agent architecture

    Google Cloud introduced Project Chrysalis, a research reference architecture for governed, self-evolving AI agents that can diagnose outages, synthesize mitigation code, assess interventions, retain operational lessons and propose changes to an agent hierarchy. Built on the Gemini platform and Agent Development Kit, it uses Model Armor, Cloud KMS, Binary Authorization, Open Policy Agent and human approval gates to constrain autonomous actions. In automated Black Friday cache-stampede demo simulations, Google reported that time to first mitigation fell from 15 to 45 minutes to 1.8 seconds, HTTP 5xx error rates fell to 0.0%, and P99 latency dropped from 519.8 milliseconds to 18.2 milliseconds. The report did not provide independent production validation.

    Google Developers Blog · 🔥 0
  15. DeepMind says Gemini 4 Argon built an interactive 3D human anatomy model

    Google DeepMind said in a post on X that Gemini 4 Argon combined three open-source datasets to build an interactive 3D visualization of human anatomy. According to the post, users can toggle different layers in motion, isolate muscles, or zoom inside a beating heart. The claim comes from a single official DeepMind post with no independent verification or technical detail on how the model produced the visualization.

    Google DeepMind (X) · 🔥 0
  16. PHP releases fixes for two critical unauthenticated RCE flaws

    PHP released PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35 to fix CVE-2026-109618 and CVE-2026-109619, two critical flaws that can enable unauthenticated remote code execution under certain conditions. The PHP project said the bugs involve memory corruption and stale state in core memory-management and error-handling paths. Simon Willison’s Weblog and The Next Web reported that AI tools assisted vulnerability research and report validation, while the PHP Foundation retained human responsibility for selecting patches and assessing compatibility costs.

    Hacker News 首页(100+ 分) · 🔥 0
  17. DeepMind: Gemini 4 Argon pipeline finds 20 candidate Cryptids in Busy Beaver 6

    Google DeepMind's Discovery team, in a project led by student researcher dsantosmarco, reported new results on the sixth Busy Beaver problem using a pipeline powered by Gemini 4 Argon that simulates raw Turing machine traces to compress recurring patterns, conjectures mathematical transition rules, runs synthetic experiments on edge cases and proves equivalences formally in Lean 4. According to Tom Zahavy's post, the pipeline produced 20 new potential Cryptids — tiny programs whose halting behavior is equivalent to deep unsolved number theory problems — and verified 33 more machines previously studied by the Busy Beaver community. The team says all proofs were open-sourced and shared in the community Discord channel and that related halting and non-halting proofs were contributed to the community GitHub repository; the counts have not been independently verified.

    Google DeepMind (X) · 🔥 0
  18. PHP2Shell: critical PHP flaws patched after AI-assisted discovery

    Wiz Research disclosed two critical vulnerabilities in core PHP, tracked as CVE-2026-109618 and CVE-2026-109619 and nicknamed PHP2Shell, involving memory disclosure and unauthenticated remote code execution; Wiz says every PHP version released in the last decade, back to 5.4, is affected. PHP released patched versions 8.5.12, 8.4.27, 8.3.36 and 8.2.35 on October 21, 2026, with both flaws rated CVSS 9.2, while OpenAI said it coordinated with PHP maintainers, Wiz, Cloudflare and others over roughly two months. OpenAI said Calif researcher Ryan Austin identified a critical flaw using GPT-5.6 Sol under its Patch the Planet initiative, with Wiz independently discovering and reporting both flaws; Calif later said Austin was the first to find and exploit the flaw, with AI compressing verification rather than supplying direction.

    Hacker News · AI(100+ 分) · 🔥 0
  19. YouMind's 'writing taste' benchmark puts Opus 5.5 first

    YouMind published an internal benchmark it calls "文感力" (writing taste) that scores how closely a model's choices in a specific context match those of an experienced author, rather than whether it merely meets a brief. The method takes recent well-received articles by known authors, removes a paragraph, turns the reasons the original works into scoring criteria and builds wrong answers that fail individual criteria, then tests models on both picking the original paragraph and writing the removed one from scattered and decoy material. YouMind says Opus 5.5 ranked first, Gemini 4 Argon was the most reliable reader and Gemini 3.8 Flash offered the best value, and that Opus 5.5 and GPT-6 Luna, previously near-tied in its tests, differed by 42 points; it also reports every model wrote longer than the original (median 1.6x) and that 58% of rewrites broke at least one of the author's trade-offs even on articles the model had read correctly. These are vendor-reported internal results without independent replication.

    即刻 · AI探索站 · 🔥 0
  20. Google diverts cloud compute to frontier AI as 2027 capacity tightens

    Google CEO Sundar Pichai said the company is shifting some compute from Google Cloud toward frontier AI research, in some cases sacrificing near-term cloud revenue. He also said customers seeking 2027 capacity face extreme difficulty because data centers take two to three years to deploy while demand for intelligent-agent workloads is rising; Google rented external compute last year to meet some customer needs.

    InfoQ 中文 AI&大模型 · 🔥 0
  21. MiniMax M3.1 Flash benchmark gains spark training-data allegations

    MiniMax M3.1 Flash earned an Artificial Analysis Intelligence Index score of 42, 13 points above M3, but researchers questioned whether the gain reflected genuine capability or memorized benchmark questions. Researcher Binkg reported that the model retained the original answer after a public AA-Omniscience question was modified, and TMTPost reproduced a similar English-language failure, although the model answered correctly in Chinese. MiniMax opened the model weights on October 23, enabling independent testing, but no request logs or training records show that the company accessed or used private evaluation questions.

    cnBeta · 🔥 0
  22. Gemini 4 Argon agents extend fusion magnetic-confinement math

    Google's Gemini 4 Argon, run through the Antigravity environment, helped researchers extend mathematical solutions tied to the magnetic fields inside experimental fusion reactors. The study by Opal Issan, Jonathan Citrin, Matt Landreman and Brendan Tracey builds on exact solutions Landreman published in September 2026 with help from OpenAI's GPT-6 Astra Pro, adding a parameter that breaks their shared geometric symmetry to give researchers more exact reference cases for checking magnetic-equilibrium software. Promising candidates reportedly emerged within hours, but the authors still had to verify the derivations; the article states this is not a fusion-energy or commercial-viability breakthrough, and its headline framing that a 59-year-old conjecture was "shattered" overstates the result.

    TechRadar · 🔥 0
  23. Arena Week 42 leaderboard: kimi-k3-max enters coding top 10

    Arena published its blind-test rankings for Week 42 of 2026 (October 12–18), with the overall top stable: gemini-4-argon-high retained first place at 1525 points and claude-opus-5.5-high was second at 1507. On the coding board, Moonshot's kimi-k3-max held tenth with 1540 points, followed by Xiaomi's mimo-v2.6-pro in eleventh at 1539. Three models newly entered the rankings this week—gemini-3.5-flash-medium, mimo-v2.6-pro and flux-2-flex—and Arena notes the boards use anonymous blind votes and ELO, reflecting user preference rather than an absolute capability test.

    IT之家 AI · 🔥 0
  24. Gemini 4 Argon trails Claude Opus 5.5 in third-party scoring

    Google released Gemini 4 Argon, the top tier of its Gemini 4 family, on September 30, and TechNews reports that Artificial Analysis third-party scoring gives Argon 53 points — level with OpenAI's GPT-6 Astra and below Anthropic Claude Opus 5.5 at 58, citing an ExaWizards analysis republished by ITmedia. Epoch AI estimates Google held about 25% of global AI compute as of the end of 2025, the largest share. SemiAnalysis founder Dylan Patel said on the Big Technology Podcast that the problem is not total compute but competition for resources between Google Cloud and DeepMind: Google's overall compute has grown yearly while the share allocated to DeepMind has fallen, a situation he summed up as "Kurian won, Hassabis lost." The report also notes Patel's forecast that DeepMind will have the least compute of the three by the end of 2026.

    科技新报 · 🔥 0
  25. Daxiao releases ACE-Brain-0.6 embodied model, claims six grounding wins over GPT-6 Astra

    Daxiao Robotics released ACE-Brain-0.6, a vision-spatial grounding-enhanced embodied foundation model built by supervised fine-tuning Qwen3.5-9B on roughly one million spatial grounding training samples. The company says it beat OpenAI's GPT-6 Astra on six of eight grounding evaluations, with a 74.53 average across seven public benchmarks versus Astra's 73.68, and scored 94.13 on its self-built RobotGrounding real-robot evaluation versus Astra's 91.10. All these evaluations were run by the vendor and have not been independently reproduced.

    品玩 实时要闻 · 🔥 0
  26. Google releases Gemini 4 Argon for complex professional work

    Google released Gemini 4 Argon on October 19, saying it is available through the Gemini API to global developers and enterprises and to Google AI subscribers in Google Antigravity, the Gemini App and AI Mode in Search. Launch pricing is $2 per million input tokens and $10 per million output tokens, with a 95% discount on cached input tokens. Google says Argon leads other frontier models on benchmarks for coding, finance, legal, cybersecurity and science, including first place on the Vals index and 87% on the ERQA embodied-reasoning benchmark; those benchmark and productivity figures are vendor-reported.

    InfoQ 中文 AI&大模型 · 🔥 0
  27. Tomasz Tunguz: Anthropic looks like a cloud, and should be valued like one

    In a blog post, venture investor Tomasz Tunguz argues that Anthropic should be assessed as a cloud business rather than only an AI model company, comparing its Q2 2026 revenue of $12 billion with AWS at $42 billion, Azure at $39 billion and Google Cloud at $25 billion. He cites 1,374% year-over-year growth, a $46 billion annual run rate and $518 billion in data-center commitments, while noting that two customers accounted for nearly 25% of 2025 revenue, per Reuters reporting from an unfiled prospectus. He also cites Ramp vendor data showing 87% of businesses buying a coding agent use Claude Code, and Artificial Analysis figures putting Claude Opus 5.5 at $5.98 per task versus $3.26 for GPT-6 Astra and $1.99 for Gemini 4 Argon.

    Tomasz Tunguz · 🔥 0
  28. KillSwitch tests whether LLMs truly understand unfamiliar code

    KillSwitch creator introduced an open-source esoteric programming language designed to challenge LLMs and coding agents through letter-count rules, hidden misleading text, noisy scripts, prompt-like phrases, and safety-sensitive constructs. The creator’s benchmark report says Claude Opus 5 ranked highest with a score of 66.8, while Gemini 4 Argon and Fable 5.1 performed poorly; however, these results come from the project creator and are not independently verified in the supplied report. The author cautions that the tasks are simple and that completing them does not establish that models genuinely understood the language.

    Hacker News · AI(100+ 分) · 🔥 0
  29. Gemini 4 Argon scores 97.5% on ARC-AGI-3

    ARC Prize reported on Oct. 16 that Google DeepMind's Gemini 4 Argon scored 97.5% on ARC-AGI-3 using the Provider Adapter Harness at a cost of $4,800, making Google the second lab after OpenAI to cross 95% on the benchmark. Under the Standard Harness the same model fell to 49.8% at low reasoning for $5,800, and API problems meant it completed only 19 games at the highest reasoning setting, so no score was published. Argon also scored 98.5% on ARC-AGI-1 and 91.3% on ARC-AGI-2, while Epoch AI's Capabilities Index puts it at 165, just behind Claude Opus 5.5 at 167 and GPT-6 Astra at 166, with weaker relative standing in math and software engineering.

    The Decoder · 🔥 0
  30. StarSkirmish Bench v0.2: GPT-6 Astra tops LLM StarCraft bot ranking

    StarSkirmish Bench v0.2 gives each large model one hour to write a Protoss bot in C++ for StarCraft: Brood War, then scores it by win rate against human-written and demo bots. GPT-6 Astra leads with 57, followed by GPT-6.1 Sol at 46, with Gemini 4 Argon and Claude Opus 5.5 tied at 39, Claude Fable 5.1 at 37, then Claude Sonnet 5.5 (26), Muse Wonder Full and Claude Haiku 5.5 (21) and GPT-6 Luna (13). Even the best model, GPT-6 Astra, wins only about 7% of games against top human-written bots Stardust, PurpleWave and BananaBrain, so the benchmark is far from saturated; practicing against strong opponents is the strongest predictor of a model's score, with Claude Fable 5.1 a clear outlier.

    Hacker News · AI(100+ 分) · 🔥 0

← PreviousNext →

Experience and discussion from the community

Share my Gemini 4 Argon experienceAsk about Gemini 4 Argon

Nobody has shared their experience with Gemini 4 Argon yet.