Gemini 4 Argon
Gemini 4 Argon, a 2026‑10‑13 general‑purpose model, continues to serve as Gemini’s action‑planning engine. A recent experiment tied to Gemini Robotics showed that Gemini 4 Argon could decompose a user‑instructed task into steps that Genie 4 simulated in real time, demonstrating integration between Gemini 4 Argon and the world model. Gemini 4 Argon is used to tie Genie 4 work to Gemini Robotics research. Gemini 4 Argon, via Genie 4, simulates Apptronik’s Apollo humanoid robot playing checkers without using a real robot. Gemini 4 Carbon will launch at the same introductory price as Argon ($2 per million input tokens and $10 per million output tokens). Gemini 4 Carbon builds on Gemini 4 Argon, which was launched two weeks ago. Gemini 4 Argon is the previous top model before Gemini 4 Carbon Gemini 4 Carbon is a slightly better version of Gemini 4 Argon. Gemini 4 Carbon is apparently twice as likely to generate code that is free of vulnerabilities as Gemini 4 Argon. Gemini 4 Carbon was released two weeks after Gemini 4 Argon Gemini 4 Carbon overtops Gemini 4 Argon in the Text Arena (1,528 vs 1,520 points) Gemini 4 Carbon generates 2.3 times as many reasoning tokens as Gemini 4 Argon Gemini 4 Carbon generates 64% more output tokens than Gemini 4 Argon Gemini 4 Carbon offers significant gains over Argon in coding, agentic tasks, and cybersecurity Google released Gemini 4 Argon during the quarter· Graphiti 生成的实体摘要
关系变化
按生效时间- Gemini 4 ArgonWILL BE EQUAL PRICE ASClaude Opus 5
On 2026‑12‑01 the Gemini 4 Argon model’s price will be equal to that of Claude Opus 5.5
- GoogleRELEASESGemini 4 Argon
Google released Gemini 4 Argon during the quarter
- Gemini 4 CarbonIMPROVES OVERGemini 4 Argon
Gemini 4 Carbon offers significant gains over Argon in coding, agentic tasks, and cybersecurity
- Gemini 4 CarbonMULTIPLES REASONING TOKENS THANGemini 4 Argon
Gemini 4 Carbon generates 2.3 times as many reasoning tokens as Gemini 4 Argon
- Gemini 4 CarbonOVERTOPGemini 4 Argon
Gemini 4 Carbon overtops Gemini 4 Argon in the Text Arena (1,528 vs 1,520 points)
- Gemini 4 CarbonMULTIPLES OUTPUT TOKENS THANGemini 4 Argon
Gemini 4 Carbon generates 64% more output tokens than Gemini 4 Argon
- Gemini 4 CarbonIS TWICE AS LIKELY TO GENERATE CODE FREE OF VULNERABILITIES ASGemini 4 Argon
Gemini 4 Carbon is apparently twice as likely to generate code that is free of vulnerabilities as Gemini 4 Argon.
- Gemini 4 CarbonIS BETTER THANGemini 4 Argon
Gemini 4 Carbon is a slightly better version of Gemini 4 Argon.
- Gemini 4 CarbonLAUNCHES AT SAME PRICE ASGemini 4 Argon
Gemini 4 Carbon will launch at the same introductory price as Argon ($2 per million input tokens and $10 per million output tokens).
- Gemini 4 CarbonBUILDS ONGemini 4 Argon
Gemini 4 Carbon builds on Gemini 4 Argon, which was launched two weeks ago.
- Gemini 4 ArgonPREVIOUS MODEL OFGemini 4 Carbon
Gemini 4 Argon is the previous top model before Gemini 4 Carbon
- Gemini 4 ArgonUSED TO SIMULATEApptronik’s Apollo humanoid
Gemini 4 Argon, via Genie 4, simulates Apptronik’s Apollo humanoid robot playing checkers without using a real robot.
- Gemini 4 ArgonASSOCIATED WITHGemini Robotics
Gemini 4 Argon is used to tie Genie 4 work to Gemini Robotics research.
- Gemini 4 ArgonBROKE DOWN TASKS FORGenie 4
Gemini 4 Argon model broke down tasks for Genie 4
- Gemini 4 ArgonPART OFGemini
Gemini 4 Argon is part of Gemini
- Gemini 4 ArgonPART OFGemini Robotics
Gemini 4 Argon model is part of Gemini Robotics
- Gemini 4 ArgonWORKS WITHGENIUS Act
Gemini 4 Argon works with Gemini Robotics
- Gemini 4 ArgonIS AGemini
Gemini 4 Argon is a Gemini model
- GPT-6.1 TerraHAS LESS SCORE THANGemini 4 Argon
GPT‑6.1 Terra has a lower Intelligence Index score (48) than Gemini 4 Argon’s score (53)
- Kimi K3.1 ProOUTPERFORMSGemini 4 Argon
Kimi K3.1 Pro outperforms Gemini 4 Argon in Artificial Analysis scoring
- WizUSEDGemini 4 Argon
Wiz used Gemini 4 Argon
- WizUSESGemini 4 Argon
Wiz uses Gemini 4 Argon for discovery and remediation of PHP2Shell
- Gemini 4 ArgonUSED BYWiz
Gemini 4 Argon was used by Wiz.
- Gemini 4 ArgonFOUND VULNERABILITY INPHP
Gemini 4 Argon found a vulnerability in PHP
- GoogleCLAIMEDGemini 4 Argon
Google claimed Gemini 4 Argon optimized speedup
- Gemini 4 ArgonRELEASED BYDeepMind
Gemini 4 Argon was released by Google DeepMind on 2026-09-30
- Gemini 4 ArgonHAS CAPABILITY INDEX GREATER THANGemini 3.8 Flash
Gemini 4 Argon has a higher capability index than Gemini 3.8 Flash
- Gemini 4 ArgonBELONGS TODeepMind
Gemini 4 Argon belongs to Google DeepMind
- Gemini 4 ArgonHAS CAPABILITY INDEX GREATER THANGemini 1.5 Pro
Gemini 4 Argon has a higher capability index than Gemini 1.5 Pro
- DeepMindCONDUCTED INTERNAL AUDIT FORGemini 4 Argon
Google DeepMind conducted an internal audit for Gemini 4 Argon
- Gemini 4 ArgonSCORES ONERQA benchmark
Gemini 4 Argon scores 87 percent on the ERQA benchmark
- GoogleRELEASEDGemini 4 Argon
Google released Gemini 4 Argon globally just days earlier.
- Gemini 4 ArgonAVAILABLE INGemini API
Gemini 4 Argon is available through the paid tier of the Gemini API.
- Gemini 4 ArgonBUILDSMuJoCo simulation
Gemini 4 Argon builds a MuJoCo simulation for training the Spot robot.
- GooglePITCHESGemini 4 Argon
Google pitches Gemini 4 Argon as the single model for the full robotics pipeline.
- Google AI UltraALLOWS ACCESS TOGemini 4 Argon
Google AI Ultra subscribers have access to Gemini 4 Argon
- Gemini 4 ArgonCONTROLSBoston Dynamics Spot robot
Gemini 4 Argon controls the Boston Dynamics Spot robot.
- Gemini 4 ArgonUSESMuJoCo simulation
Gemini 4 Argon uses MuJoCo simulation to create a virtual copy of the bin
- Gemini 4 ArgonDIRECTSGemini Omni
Gemini 4 Argon directs the Gemini Omni video model to generate a training video.
- Gemini 4 ArgonSCORES ONYounite AI
Gemini 4 Argon scored 87 percent on the ERQA benchmark.
关系是 Graphiti 从报道中提取的事实陈述,不是原文引述;失效时间表示后来的报道显示这条关系已不成立。
相关事件 12
- Google Personal Intelligence 功能推送至 Android Auto6 篇相关报道 · 2026-10-29
- Google expands AI Search ads globally1 篇相关报道 · 2026-10-29
- Google发布Genie 4世界模型3 篇相关报道 · 2026-10-28
- OpenAI发布GPT-6.1 Terra并降低API价格1 篇相关报道 · 2026-10-27
- Kimi K3.1 Pro 上线 Moonshot API1 篇相关报道 · 2026-10-24
- EU cyber team warns websites to go offline due to PHP vulnerabilities2 篇相关报道 · 2026-10-22
- Patch the Planet:安全研究人员协同修复开源关键漏洞2 篇相关报道 · 2026-10-22
- 四个 Gemini 4 Argon 代理首次全自动击败 Minecraft1 篇相关报道 · 2026-10-18
- Goodfire essay challenges polysemanticity in neural networks1 篇相关报道 · 2026-10-16
- Google Deepmind报告:自改进AI仍遥遥无期2 篇相关报道 · 2026-10-16
- Google Gemini 4 Argon 向付费用户推出,高级功能受订阅等级限制6 篇相关报道 · 2026-10-16
- Modal 估值突破 157.5 亿美元1 篇相关报道 · 2026-10-15
来源报道 30
- Gemini 4 Carbon 登顶基准,首次对外开放给网络安全人员Latent Space · 2026-10-29
- Google 将 AI 搜索广告扩展至国际市场TechCrunch · AI · 2026-10-29
- Google 发布 Gemini 4 Carbon,在网络安全基准测试中位居第一The Verge · AI · 2026-10-29
- Google Deepmind 发布 Gemini 4 Carbon,两周后在 Intelligence Index 中登顶The Decoder · 2026-10-29
- Google announces Gemini 4 Carbon, its latest frontier AI model that you can't useArs Technica · AI · 2026-10-29
- Gemini 4 Carbon 登顶 AI 排行榜,但普通用户暂无购买入口BeInCrypto · 2026-10-29
- Gemini 4 Carbon:Google DeepMind 下一代前沿模型Google DeepMind · 一手 · 2026-10-29
- Google Deepmind发布Genie 4,具备实时世界模拟和对话能力The Decoder · 2026-10-28
- Google 发布 Genie 4 世界模型,增加音频、更逼真世界和实时转向TechCrunch · AI · 2026-10-28
- DeepMind 发布 Genie 4:原生音频交互的世界模型Google DeepMind · 一手 · 2026-10-28
- OpenAI 以价格而非基准测试推出 GPT-6.1 TerraBeInCrypto · 2026-10-27
- Kimi K3.1 Pro 发布,首次开放权重前沿模型Simon Willison · 2026-10-24
- OpenAI协调多方在两个月内发现PHP核心库PHP2Shell漏洞,影响70%网站Latent Space · 2026-10-22
- OpenAI GPT-5.6 Sol与Google Gemini 4 Argon发现影响数百万网站的PHP关键漏洞The Decoder · 2026-10-22
- PHP 8.x 系列发布紧急安全更新,修复两处远程代码执行漏洞Simon Willison · 2026-10-22
- PHP Armageddon:OpenAI与Google Wiz发现古老漏洞,互联网警报拉响TechCrunch · AI · 2026-10-22
- 四个 Gemini 4 Argon 代理在无人工干预下击败 MinecraftThe Decoder · 2026-10-18
- AINews 周报:Muse Spark Open 发布,GLM-5.3 等模型更新Latent Space · 2026-10-16
- Google称AI尚未构建自身——尽管它撰写了75%的代码Decrypt · 2026-10-16
- Google发布 Gemini 4 Argon:面向机器人的通用 AI 模型The Decoder · 2026-10-16
- Modal 突破 1B ARR 成为 NYC 新 AI Infra Decacorn,估值 15.75BLatent Space · 2026-10-15
- Google 称自进化 AI 仍遥遥无期The Decoder · 2026-10-15
- AI 竞赛重燃:Google 和 Meta 同日发布前沿模型The Verge · AI · 2026-10-15
- AINews:Gemini 4 Argon 全球发布及 AI Engineer NYC 事件综述Latent Space · 2026-10-14
- llm-gemini 0.35 支持 Gemini 4 Argon 模型Simon Willison · 2026-10-14
- Google DeepMind 发布 Gemini 4 Argon:负责任与安全的深度工作模型Google DeepMind · 一手 · 2026-10-14
- Gemini 4 Argon 全球发布,成为 Google 最强旗舰模型The Decoder · 2026-10-14
- Gemini 4 Argon:专为解决最难问题而构建Google DeepMind · 一手 · 2026-10-14
- Gemini 4 Argon 付费用户开始推出,仅限付费用户且高级功能受限Ars Technica · AI · 2026-10-14
- Gemini 4 Argon:Built to solve your hardest problemsGoogle AI 官方博客 · 一手 · 2026-10-14