--- title: Gemini 3.8 Flash is Google's third Flash drop in six... description: Google released Gemini 3.8 Flash, its third Flash model in six weeks, now generally available across Gemini, Google AI Studio, and the API. It scores 71% on... canonical: https://daily.dev/posts/gemini-3-8-flash-is-google-s-third-flash-drop-in-six-weeks-and-it-s-gunning-for-claude-wyexjkmxw twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Gemini 3.8 Flash is Google's third Flash drop in six weeks, and it's gunning for Claude | daily.dev og:description: Google released Gemini 3.8 Flash, its third Flash model in six weeks, now generally available across Gemini, Google AI Studio, and the API. It scores 71% on... og:url: https://daily.dev/posts/gemini-3-8-flash-is-google-s-third-flash-drop-in-six-weeks-and-it-s-gunning-for-claude-wyexjkmxw og:image: https://api.daily.dev/og/posts/WyexJKmxW.png og:image:alt: Gemini 3.8 Flash is Google's third Flash drop in six weeks, and it's gunning for Claude og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Gemini 3.8 Flash is Google's third Flash drop in six weeks, and it's gunning for Claude **[Trends](https://daily.dev/sources/trends)** · 2 min read · 15 upvotes · 1 comments ## Summary Google released Gemini 3.8 Flash, its third Flash model in six weeks, now generally available across Gemini, Google AI Studio, and the API. It scores 71% on DeepSWE 1.1 versus Claude Opus 5's 74%, at a much lower price: $0.75/1M input tokens and $3.75/1M output tokens (including thinking tokens) through end of 2026, doubling January 1, 2027. It's now the default model in Antigravity and Managed Agents, and was spotted in production before the official announcement. Google's Phil Schmid frames the rapid cadence as focused on agentic autonomy and long-horizon tasks, while Primeagen sarcastically urged people to check the model card before hype takes over. ## Content Three Gemini Flash releases in six weeks. 3.6, 3.7, 3.8. As @trikcode put it: "how are we supposed to keep up with this." Gemini 3.8 Flash is GA across Google AI Studio, the Gemini API, Cursor, GitHub Copilot, JetBrains, Vercel AI Gateway, and GCP Agent Studio. Pricing holds at $0.75/$3.75 per million input/output tokens through December 31, then doubles. The model is designed to take smaller steps and verify its own work more often — which means higher token usage on complex tasks, but Google's position is that those tokens are doing real work. The benchmark numbers are genuinely interesting. 73.8% on DeepSWE 1.1, beating Gemini Astra's 73.3%. 54.9% on HLE-Verified versus Claude Opus 5's 54.4%. 89.4% on Terminal-Bench 2.1. 69.9% on the Cursor bench at $2.38 per task. @rohanpaul_ai flagged the Harvey Legal Agent score — 10.0% versus Opus 5's 6.7% — which sounds low until you understand the benchmark gives credit only when every single criterion passes across complex legal workflows. Not everyone is impressed. @emollick, who had early access, called it "a very good Flash model, but not equivalent to a frontier model." @ThePrimeagen's take was more pointed: the model is fast, yes — fast at reading the same two files twenty times in a row. @jerryjliu0 benchmarked it on document parsing and found it roughly flat with 3.7 and 3.6 Flash on overall scores, with some regressions on charts and semantic formatting. The more interesting discourse is about what's *not* shipping. Google also dropped Flash Cyber — a vulnerability-detection variant claiming 2.6x better patch accuracy than larger commercial models — but it's gated to roughly 650 trusted partners and governments through something called the Fairwind Program. @theo's reaction: "We're never getting a new Gemini Pro model huh." @Hesamation went further: "Google cooked so hard with 3.8 Flash that they can't ship a Pro that justifies the higher price." That's the real tension here. Flash keeps getting better, fast, cheap, and genuinely competitive on agentic coding tasks. The Pro line has been quiet since early 2026. At some point the question stops being "is Flash good enough" and starts being "what is Pro even for." ## Questions this post answers ### What is the pricing for Gemini 3.8 Flash and when does it change? Gemini 3.8 Flash is priced at $0.75 per million input tokens and $3.75 per million output tokens, holding through December 31 before the rate doubles. It is generally available across Google AI Studio, the Gemini API, Cursor, GitHub Copilot, JetBrains, Vercel AI Gateway, and GCP Agent Studio. _Track shifting LLM API pricing like this on daily.dev before locking in a model for production._ ### How does Gemini 3.8 Flash compare to Claude Opus 5 on benchmarks? Gemini 3.8 Flash scored 54.9% on HLE-Verified versus Claude Opus 5's 54.4%, and 10.0% on the Harvey Legal Agent benchmark versus Opus 5's 6.7%, though that legal benchmark only credits answers passing every criterion across complex workflows. It also hit 73.8% on DeepSWE 1.1 and 89.4% on Terminal-Bench 2.1. _Developers picking between Gemini and Claude for coding agents can follow benchmark comparisons on daily.dev._ ### What is Google's Flash Cyber and who can access it? Flash Cyber is a vulnerability-detection variant of Gemini Flash that Google claims achieves 2.6x better patch accuracy than larger commercial models. Access is restricted to roughly 650 trusted partners and governments through the Fairwind Program, so it isn't publicly available like the standard Flash models. _Security teams weighing gated AI vulnerability-detection tools can keep up with releases like Flash Cyber on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 1 discussion and 22 comments across x (as of 2026-09-13). **TL;DR:** Reaction is dominated by fatigue and skepticism about the rapid pace of Flash releases, with repeated calls for a proper Pro model instead, and little enthusiasm for Flash itself. **Sentiment:** 5% positive · 25% mixed · 70% skeptical **The case for** - A few note that despite rapid iteration, output quality hasn't degraded. **The pushback** - Frequent versioning makes it hard to track which release is actually current. - Frustration that prompts and behavior shift unpredictably between point releases. - Flash models are seen as underwhelming compared to hoped-for Pro releases. - Some feel the brand hasn't achieved the recognition of competing model families. - A sense that constant baseline changes undermine building durable products on top of the model. **By community** - x (skeptical): Replies mostly express fatigue with rapid, confusingly named releases and disappointment that a new Pro model hasn't shipped instead. **Open questions** - When will a genuinely new Pro model ship instead of another Flash iteration? **Highlights** > @trikcode you can spend months building around a model and suddenly the baseline moves underneath you at this point the moat can't just be the model. it has to be what you build around it. > — [daeneilsup on x](https://x.com/daeneilsup/status/2096905765034578197) > @trikcode The exhausting part isn't the versions, it's that each one quietly changes how your prompts behave. Same code, different output. > — [Kishan\_Devani\_ on x](https://x.com/Kishan_Devani_/status/2096897139737796990) > @trikcode six weeks is not that fast for a point release but the naming is doing a lot of work here > — [0xTreff on x](https://x.com/0xTreff/status/2096896353230066166) > @trikcode All flash models from gemini is flop , expecting pro models will not disappoint . waiting for gemini 3.6 pro model > — [naveen\_saradhi\_ on x · 1 comments](https://x.com/naveen_saradhi_/status/2096898749171986454) > @trikcode Stopped keeping up on purpose. I build with whatever's default and ship. Chasing point releases felt like tuning a car I can't drive. My last app shipped on whatever was default that week. > — [LucaCaponeX on x](https://x.com/LucaCaponeX/status/2096906758920761506) **Source threads** - [x](https://x.com/trikcode/status/2096895839780159705) · 0 points · 22 comments ## Community discussion Top comments from developers on daily.dev. **@leonidbugaev** · 1 upvotes > It is actually quite good. Worth noting that currently there is a 50% price. That's why it's also very good. > > Also, fun story: a few days ago, we tried to migrate to the OpenVEX models for the internal AI bot, a quite complex one, and GLM 5.3 failed a lot, and was actually more expensive (spends more tokens). Before, we were using the Gemini 3.1 Pro. After a few days, we have migrated to the 3.8 Flash, and so far, so good: fast and quality. --- Tags: [#google](https://daily.dev/tags/google), [#llm](https://daily.dev/tags/llm), [#claude](https://daily.dev/tags/claude), [#google-gemini](https://daily.dev/tags/google-gemini), [#agentic-ai](https://daily.dev/tags/agentic-ai) [View this post on daily.dev](https://daily.dev/posts/gemini-3-8-flash-is-google-s-third-flash-drop-in-six-weeks-and-it-s-gunning-for-claude-wyexjkmxw) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Gemini 3.8 Flash is Google's third Flash drop in six weeks, and it's gunning for Claude","url":"https://daily.dev/posts/gemini-3-8-flash-is-google-s-third-flash-drop-in-six-weeks-and-it-s-gunning-for-claude-wyexjkmxw","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/gemini-3-8-flash-is-google-s-third-flash-drop-in-six-weeks-and-it-s-gunning-for-claude-wyexjkmxw"},"datePublished":"2026-09-02T15:43:28.472Z","dateModified":"2026-09-13T19:33:14.688Z","description":"Google released Gemini 3.8 Flash, its third Flash model in six weeks, now generally available across Gemini, Google AI Studio, and the API. It scores 71% on...","image":"https://pbs.twimg.com/media/HROPESqbYAAl7IM.jpg","thumbnailUrl":"https://pbs.twimg.com/media/HROPESqbYAAl7IM.jpg","isAccessibleForFree":true,"articleSection":"Trends","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Trends","logo":"https://media.daily.dev/image/upload/s--ZfSp3asX--/f_auto,q_auto/v1780996004/logos/trends?_a=BAMAMiWQ0","url":"https://daily.dev/sources/trends"},"commentCount":1,"discussionUrl":"https://daily.dev/posts/gemini-3-8-flash-is-google-s-third-flash-drop-in-six-weeks-and-it-s-gunning-for-claude-wyexjkmxw","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":15},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":1}],"keywords":"google,llm,claude,google-gemini,agentic-ai","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Trends","item":"https://daily.dev/sources/trends"},{"@type":"ListItem","position":3,"name":"Gemini 3.8 Flash is Google's third Flash drop in six weeks, and it's gunning for Claude"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/gemini-3-8-flash-is-google-s-third-flash-drop-in-six-weeks-and-it-s-gunning-for-claude-wyexjkmxw","comment":[{"@type":"Comment","text":"It is actually quite good. Worth noting that currently there is a 50% price. That’s why it’s also very good.\nAlso, fun story: a few days ago, we tried to migrate to the OpenVEX models for the internal AI bot, a quite complex one, and GLM 5.3 failed a lot, and was actually more expensive (spends more tokens). Before, we were using the Gemini 3.1 Pro. After a few days, we have migrated to the 3.8 Flash, and so far, so good: fast and quality.","datePublished":"2026-09-03T14:08:25.705Z","url":"https://daily.dev/posts/WyexJKmxW#c-XB8oQ1bOk","author":{"@type":"Person","name":"Leonid Bugaev","url":"https://daily.dev/leonidbugaev","image":"https://lh3.googleusercontent.com/a/ACg8ocLhMgurwTmJElWaH9A8Ju8cHvZxgMCDt009jEYkmFCyUGAoaGSw=s96-c"},"interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/gemini-3-8-flash-is-google-s-third-flash-drop-in-six-weeks-and-it-s-gunning-for-claude-wyexjkmxw#faq","mainEntity":[{"@type":"Question","name":"What is the pricing for Gemini 3.8 Flash and when does it change?","acceptedAnswer":{"@type":"Answer","text":"Gemini 3.8 Flash is priced at $0.75 per million input tokens and $3.75 per million output tokens, holding through December 31 before the rate doubles. It is generally available across Google AI Studio, the Gemini API, Cursor, GitHub Copilot, JetBrains, Vercel AI Gateway, and GCP Agent Studio. Track shifting LLM API pricing like this on daily.dev before locking in a model for production."}},{"@type":"Question","name":"How does Gemini 3.8 Flash compare to Claude Opus 5 on benchmarks?","acceptedAnswer":{"@type":"Answer","text":"Gemini 3.8 Flash scored 54.9% on HLE-Verified versus Claude Opus 5's 54.4%, and 10.0% on the Harvey Legal Agent benchmark versus Opus 5's 6.7%, though that legal benchmark only credits answers passing every criterion across complex workflows. It also hit 73.8% on DeepSWE 1.1 and 89.4% on Terminal-Bench 2.1. Developers picking between Gemini and Claude for coding agents can follow benchmark comparisons on daily.dev."}},{"@type":"Question","name":"What is Google's Flash Cyber and who can access it?","acceptedAnswer":{"@type":"Answer","text":"Flash Cyber is a vulnerability-detection variant of Gemini Flash that Google claims achieves 2.6x better patch accuracy than larger commercial models. Access is restricted to roughly 650 trusted partners and governments through the Fairwind Program, so it isn't publicly available like the standard Flash models. Security teams weighing gated AI vulnerability-detection tools can keep up with releases like Flash Cyber on daily.dev."}}]} ```