{"slug": "google-is-internally-testing-out-gemini-3-8-flash-reports", "title": "Google Is Internally Testing Out Gemini 3.8 Flash: Reports", "summary": "Google is internally testing Gemini 3.8 Flash, the presumed successor to Gemini 3.7 Flash, according to Business Insider and AI leakers on X, with a possible public release in September. A Tencent research paper published August 10 listed 'Gemini 3.8 Flash' as an LLM judge, suggesting external access before 3.7 Flash's launch. Google has not confirmed the model's existence, but the Flash line has seen rapid releases: 3.6 Flash on July 21 and 3.7 Flash roughly three weeks later, with CEO Sundar Pichai calling 3.7 Flash the company's fastest-growing model ever.", "body_md": "Google’s Gemini 3.5 Pro appears cancelled, and Gemini 4.0 could be a while away, but the company is still iterating on its older Flash models.\n\nAccording to reports by Business Insider, Google is already running an internal version of Gemini 3.8 Flash, the presumed successor to [Gemini 3.7 Flash](https://officechai.com/ai/gemini-3-7-flash-benchmarks/), which itself only became publicly available a couple of weeks ago.\n\nThe information has also been corroborated on posts by AI leakers on X, with accounts including 0x0SojalSec and Salio claiming that the model has been deployed internally at Google for months, going through what they describe as extensive A/B testing and partner feedback cycles. One post claims the model is being refined specifically for better tool execution and reduced hallucinations, both areas Google has been trying to shore up as it pushes Flash models deeper into agentic workflows. A public release, if these leaks are accurate, could land sometime in September.\n\nThere’s also a more concrete data point buried in all this speculation. A Tencent research paper published on August 10, three days before Gemini 3.7 Flash’s official launch, reportedly listed “Gemini 3.8 Flash” alongside GPT-5.5 and Claude Opus 4.8 as one of three LLM judges used in its evaluation setup. That’s the kind of detail that’s harder to wave away as pure rumor, since it would mean external researchers had access to a model under that name before Google had even shipped 3.7 Flash to the public.\n\nNone of this comes from Google itself. There’s no model card, no API listing, and no confirmation from anyone at DeepMind that a model called Gemini 3.8 Flash exists in any official capacity. Given how quickly the Flash line has been moving lately, though, the claim isn’t exactly far-fetched.\n\n## Flash Has Become Google’s Real Release Cadence\n\nWhatever ends up happening with 3.8 Flash, the pattern behind it is now well established. Google shipped [Gemini 3.6 Flash](https://officechai.com/ai/googles-gemini-flash-3-6-spotted-in-antigravity-could-launch-soon/) on July 21 as a stopgap after Gemini 3.5 Pro [missed ](https://officechai.com/ai/google-has-silently-canceled-gemini-3-5-pro-says-semi-analysis-report/)its promised launch window, then followed it up with 3.7 Flash roughly three weeks later. Both releases were framed internally as algorithmic improvements rather than full retrains, and Logan Kilpatrick, who leads product for Google’s AI Studio, has said as much publicly when discussing the jump from 3.6 to 3.7.\n\nThat release has performed well by Google’s own account. Sundar Pichai called it the [company’s fastest-growing model ever](https://officechai.com/ai/google-says-that-gemini-3-7-flash-is-its-fastest-growing-model-ever/) in its first week, and it’s now running inside Search and the Gemini app rather than staying confined to the API. On benchmarks, the model has done more than hold its own too, topping Artificial Analysis’s AnalystAgent benchmark for spreadsheet and document work ahead of larger, pricier models.\n\nIf Gemini 3.8 Flash is real and does land in September, it would mean Google went from 3.6 to 3.7 to 3.8 in under two months. That’s an unusually tight cadence for any model family, let alone one that Google is positioning as competitive with frontier-tier systems.\n\n## The Bigger Picture: A Pro-Sized Hole\n\nThe reason Flash has taken on this outsized importance traces back to what happened with Gemini 3.5 Pro. Sundar Pichai announced it on stage at Google I/O in May, told developers it would arrive “next month,” and then months passed with no release. Google eventually released updates to the Flash line instead, and a SemiAnalysis note reported that the company has quietly shelved 3.5 Pro altogether, choosing to skip straight to [Gemini 4](https://officechai.com/ai/gemini-3-5-pro-is-testing-with-partners-have-begun-pre-training-run-for-gemini-4-pro-google/) rather than ship a model that leadership felt wouldn’t move the needle.\n\nGoogle has never confirmed the cancellation outright, and continues to say the model is being tested with partners. But the practical result is the same either way: there is currently no supported Gemini Pro model newer than what was available before I/O, and all of Google’s public momentum this year has come from the Flash line instead. Gemini 4 itself is also still a ways off, with Google having only confirmed in July that it had begun what it called its most ambitious pre-training run yet, with no release window attached.\n\nThat leaves Flash carrying a lot of weight it wasn’t originally built for. It was designed as the fast, cheap, high-throughput option for developers running large-scale or latency-sensitive workloads, not as Google’s primary showcase for capability gains. With Pro effectively paused and Gemini 4 still deep in training, Flash has become the only lever Google has left to pull in the near term, and the frequency of these releases reflects that.", "url": "https://wpnews.pro/news/google-is-internally-testing-out-gemini-3-8-flash-reports", "canonical_source": "https://officechai.com/ai/google-is-internally-testing-out-gemini-3-8-flash-reports/", "published_at": "2026-08-28 12:40:07+00:00", "updated_at": "2026-08-28 14:49:13.661875+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-products"], "entities": ["Google", "Gemini 3.8 Flash", "Gemini 3.7 Flash", "Business Insider", "Tencent", "Sundar Pichai", "Logan Kilpatrick", "DeepMind"], "alternates": {"html": "https://wpnews.pro/news/google-is-internally-testing-out-gemini-3-8-flash-reports", "markdown": "https://wpnews.pro/news/google-is-internally-testing-out-gemini-3-8-flash-reports.md", "text": "https://wpnews.pro/news/google-is-internally-testing-out-gemini-3-8-flash-reports.txt", "jsonld": "https://wpnews.pro/news/google-is-internally-testing-out-gemini-3-8-flash-reports.jsonld"}}