Gemini 3.8 Flash

The AIWIZ article, dated September 6, 2026, confirms Google released Gemini 3.8 Flash on September 2, 2026, with introductory API pricing running until December 31. Google's developer guidance states the model can take smaller reasoning steps, use tools repeatedly, and check its progress, which can mean consuming more Independent testing cited from Artificial Analysis gave 3.8 Flash an Intelligence Index score of 59, compared with 56 for 3.7 Flash. The article frames the model as suited for complex reasoning, coding, and multi-step tasks, following 3.7 Flash with improvements Google says strengthen the ability to complete demanding

Xpert.Digital reports on the September 2, 2026 launch of Gemini 3.8 Flash and the restricted Gemini 3.8 Flash Cyber variant, which is positioned for detecting security vulnerabilities in program code and automatically generating patches. The piece attributes Cyber's development to Tulsee Doshi, Senior Director of Produ The article highlights a 'cost trap' arising from Gemini 3.8 Flash's increased internal token consumption through expanded reasoning and iterative tool use: while the per-token introductory price matches 3.7 Flash at $0.75 input and $3.75 output per million tokens, real-world task costs rise meaningfully. Restricted ac

Tech-Insider reports that Google DeepMind shipped Gemini 3.8 Flash and Gemini 3.8 Flash Cyber on September 2, 2026, with the Cyber variant gated behind a new Fairwind Program that mirrors the restricted rollout used for the earlier Gemini 3.5 Flash Cyber. Gemini 3.8 Flash reached general availability the same day throu The article quotes Google DeepMind's official X post describing 3.8 Flash as "our most intelligent model yet with significant gains from 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning," and CTO Koray Kavukcuoglu confirming the launch timing. Both models share the same underlying core, wi

Gemini 3.8 Flash shipped under the model ID gemini-3.8-flash and is pitched at long-horizon coding and autonomous agents rather than chat, per the Eesel AI review quoting Google's model card. Specs are unchanged from the prior generation: text, image, audio, video, and PDF inputs; text output; a 1M-token context window The review highlights a model-card sentence that reframes the release: 3.8 Flash works harder on complex tasks, running extra reasoning steps and calling tools iteratively at the same speed and low cost as 3.7 Flash. Gemini 3.8 Flash Cyber is introduced as a security variant with deliberately looser mitigations, gated

Google shipped Gemini 3.8 Flash on 2 September 2026, its third Flash-tier model in six weeks following 3.6 Flash and 3.7 Flash. The model accepts up to 1,048,576 input tokens and generates up to 65,536 output tokens—the same context window dimensions as 3.7 Flash—and handles text, images, audio, video, and PDF input na The article clarifies that Gemini 3.8 Flash is not built on a new base model but reuses the 3.7 Flash architecture, reconfiguring it to burn more thinking tokens per query, which raises reasoning latency and output token spend versus 3.7 Flash at the same prompt. Google explicitly recommends developers prioritizing cos

Vellum's September 3, 2026 benchmark walkthrough confirms Google released Gemini 3.8 Flash and Gemini 3.8 Flash Cyber on September 2, 2026, framing the pair as "two variants of one foundational model," a general-purpose Flash for everyone and a Cyber twin restricted to vetted defenders through the new Fairwind Program. The excerpt highlights an important caveat: 3.8 Flash "works harder," executing extra reasoning steps and calling tools iteratively at higher effort levels, which means it can spend more tokens and partially offset the headline price advantage. The piece also relays Google's claim that the coding and reasoning gains ac

Local AI Zone reconstructs the technical story of Google DeepMind's September 2, 2026 release of Gemini 3.8 Flash, noting it is the fourth Flash-tier model in under four months and self-described as Google's "best reasoning and coding model yet" at the price and speed of its three-week-old predecessor. The release pair The engineering story is distinctive: Google did not grow the model. The model card confirms Gemini 3.8 Flash is trained on top of Gemini 3.7 Flash's architecture and weights, but was taught to work harder through more reasoning steps, iterative tool calls, and self-verification, at the cost of roughly 40% more tokens

The Register reports that Google announced Gemini 3.8 Flash on September 2, 2026, describing it as Google's fourth Flash model in as many months and a return to the frontier model conversation after months of slower Pro-tier movement. Google senior director of product management Tulsee Doshi and Gemini Security Lead Ra The model reportedly tops the DeepSWE v1.1 long-horizon software engineering leaderboard, outperforming most larger frontier models at the $0.75/$3.75 per million token introductory rate that doubles to $1.50/$7.50 in January 2027. The article also notes leadership context, including Demis Hassabis stepping down as Dee

Google released Gemini 3.8 Flash on September 2, 2026, positioning it as the company's most intelligent Flash-tier model and its third Flash release in six weeks. According to DataCamp's technical review, the model is engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows, Benchmark gains over 3.7 Flash are reported as significant in coding and tool use: Gemini 3.8 Flash scores 90.8% on Terminal-Bench 2.1 (up from 81.6% for 3.7 Flash), 54.9% on HLE-Verified, and outperforms most larger frontier models on DeepSWE v1.1, though Humanity's Last Exam stayed essentially flat at 45.4% versus 45

Ars Technica reports that Google announced Gemini 3.8 Flash on September 2, 2026, its third Flash-variant release in just six weeks, in a move that further reduces the likelihood of the previously promised Gemini 3.5 Pro. The standard model is described as a "workhorse" suitable for agentic tasks and software developme API access is available at an introductory $0.75 per million input tokens and $3.75 per million output tokens through year-end, with the regular price set at $1.50/$7.50. Google published benchmark numbers showing Gemini 3.8 Flash at the top of the DeepSWE leaderboard for complex software engineering problems and showi

Google's Gemini 3.8 Flash launched on September 2, 2026, three weeks after Gemini 3.7 Flash and one day after Anthropic shipped Claude Fable 5.1, according to Coursiv. The model offers a 1M-token context window with a 64K-token output ceiling, accepts text, images, audio, and video, and exposes customizable effort leve Availability spans the Gemini app for AI Pro and Ultra subscribers, AI Mode, Gemini for Google Sheets, the Antigravity coding environment, Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform. Promotional pricing is $0.75 per million input tokens and $3.75 per million output tokens through the end