[{"data":1,"prerenderedAt":31},["ShallowReactive",2],{"nr-en-google-gemini-3-8-flash-launch":3},{"slug":4,"title":5,"dek":6,"date":7,"time":8,"publishedAt":9,"updated":10,"updatedAt":10,"dateFmt":11,"updatedFmt":10,"kind":12,"tier":13,"author":14,"authorName":15,"topics":16,"tracker":23,"trackerLabel":24,"headlineStat":25,"image":26,"ogImage":27,"imageAlt":5,"csv":10,"minutes":28,"words":29,"html":30},"google-gemini-3-8-flash-launch","Google Releases Gemini 3.8 Flash – Third Budget Model in Six Weeks","Google is accelerating its model release cycle dramatically: Following 3.7 Flash, the company now launches 3.8 Flash with improved coding and a new cybersecurity variant. Frontier models remain scarce.","2026-09-03","04:25","2026-09-03T04:25:00+02:00","","September 3, 2026","news","standard","ideal-syka","Ideal Syka",[17,18,19,20,21,22],"Gemini","Language Models","Google","AI Pricing","Coding","Cybersecurity","\u002Fki-preis","Token Pricing & Model Updates","Third Flash model in six weeks","\u002Fnewsroom\u002Fimg\u002Fgoogle-gemini-3-8-flash-launch.webp","\u002Fog-nr\u002Fgoogle-gemini-3-8-flash-launch.en.png",2,497,"\u003Cp>Google has released \u003Cstrong>Gemini 3.8 Flash\u003C\u002Fstrong> today – the third Flash model in just six weeks. The company is doubling down on budget variants while larger frontier models like the promised Gemini 3.5 Pro remain absent. According to Ars Technica, Google hasn't shipped a frontier-level model since early 2026.\u003C\u002Fp>\n\u003Ch2>Key Facts\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>\u003Cstrong>Two variants\u003C\u002Fstrong>: Standard Gemini 3.8 Flash (for general tasks and software development) and \u003Cstrong>Gemini 3.8 Flash Cyber\u003C\u002Fstrong> (specialized for vulnerability detection)\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Introductory pricing through year-end\u003C\u002Fstrong>: \u003Cstrong>$0.75 per million input tokens\u003C\u002Fstrong> and \u003Cstrong>$3.75 per million output tokens\u003C\u002Fstrong> (regular pricing $1.50\u002F$7.50)\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Top coding performance\u003C\u002Fstrong>: Leads the DeepSWE leaderboard and outperforms larger, more expensive models\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Cybersecurity gains\u003C\u002Fstrong>: Chrome team reports \u003Cstrong>2.6x higher patch accuracy\u003C\u002Fstrong> versus predecessor; Cloud team found critical vulnerability in two hours\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2>Marginal Gains, Faster Iteration\u003C\u002Fh2>\n\u003Cp>Performance improvements over Gemini 3.7 Flash, released a couple of weeks ago, are marginal across most benchmarks – with one major exception: \u003Cstrong>coding evaluations\u003C\u002Fstrong> show significantly larger gains. Gemini 3.8 Flash now leads the DeepSWE leaderboard, which measures how well a model solves complex software engineering problems – and it does so at lower cost than competitors.\u003C\u002Fp>\n\u003Cp>Google reportedly delayed Gemini 3.5 Pro because its coding performance couldn't match other models. If these new benchmark numbers reflect reality, even Google's budget Flash variants now compete with market leaders.\u003C\u002Fp>\n\u003Ch2>Weak Spot: Computer Use\u003C\u002Fh2>\n\u003Cp>One area remains challenging: \u003Cstrong>agentic computer use\u003C\u002Fstrong>. While Gemini 3.8 Flash improves over 3.7 Flash in the OSWorld-2.0 test, it trails the market leader Claude Opus significantly. GPT also underperforms in this benchmark.\u003C\u002Fp>\n\u003Cdiv class=\"tbl-scroll\">\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Aspect\u003C\u002Fth>\n\u003Cth>Status\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd>Coding Performance\u003C\u002Ftd>\n\u003Ctd>✓ Market-leading\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Computer Use\u003C\u002Ftd>\n\u003Ctd>✗ Well behind Opus\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Availability\u003C\u002Ftd>\n\u003Ctd>Today (Standard), limited to testers & governments (Cyber)\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Access\u003C\u002Ftd>\n\u003Ctd>Pro\u002FUltra subscription in Gemini App or AI Studio\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Ch2>Cybersecurity Variant for Specialists\u003C\u002Fh2>\n\u003Cp>Gemini 3.8 Flash Cyber replaces the 3.5 version and is increasingly important for niche fields. Google reports "substantial improvements" in internal testing: the model identifies more vulnerabilities and generates working patches more frequently. Partners like \u003Cstrong>Wiz\u003C\u002Fstrong> and \u003Cstrong>Palo Alto Networks\u003C\u002Fstrong> confirm the performance gains. The Cyber variant is currently available only to trusted testers and governments.\u003C\u002Fp>\n\u003Ch2>What This Means for You\u003C\u002Fh2>\n\u003Cp>Google's aggressive Flash release cadence signals two things: First, the company is deliberately betting on \u003Cstrong>rapid iteration in the budget segment\u003C\u002Fstrong> rather than rare frontier breakthroughs. Second, Google is using introductory pricing because competitors like Anthropic have slashed token costs, and Google needs to keep developers and enterprises engaged. For German software teams and tech companies, this could mean: AI integration costs continue to drop, but the model landscape grows more fragmented. If you're working on coding tasks, 3.8 Flash deserves a test. If you're waiting for frontier models, patience remains required.\u003C\u002Fp>\n\u003Ch2>Sources\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>\u003Ca href=\"https:\u002F\u002Farstechnica.com\u002Fai\u002F2026\u002F09\u002Fgoogle-releases-gemini-3-8-flash-its-third-flash-model-in-six-weeks\u002F\">Ars Technica\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"https:\u002F\u002Fthenextweb.com\u002Fnews\u002Fgemini-3-8-flash-cyber-release-cadence-eu-ai-act-gpai-article-53\">The Next Web\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"https:\u002F\u002Fwww.thurrott.com\u002Fa-i\u002F340992\u002Fgoogle-releases-gemini-3-8-flash-and-cyber-variant\">Thurrott.com\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"https:\u002F\u002Fthe-decoder.de\u002Fgoogles-gemini-flash-3-8-arbeitet-haerter-als-der-gerade-mal-drei-wochen-alte-vorgaenger-3-7-flash\u002F\">The Decoder (DE)\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"https:\u002F\u002Fthe-decoder.com\u002Fgemini-3-8-flash-is-googles-third-budget-model-in-six-weeks-while-frontier-models-remain-mia\u002F\">The Decoder\u003C\u002Fa>\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>\u003Cem>Editorially owned by \u003Ca href=\"\u002Fen\u002Fautor\u002Fideal-syka\">Ideal Syka\u003C\u002Fa>. Sources and method: \u003Ca href=\"\u002Fen\u002Fredaktion\">Newsroom & method\u003C\u002Fa>. Tips and corrections: \u003Ca href=\"mailto:ai@i6eal.de\">ai@i6eal.de\u003C\u002Fa>.\u003C\u002Fem>\u003C\u002Fp>\n",1788411984629]