Gemini 3.8 Flash

Gemini 3.8 Flash — explained in detail

Google released Gemini 3.8 Flash on 2 September 2026 as its most intelligent workhorse model to date. It arrives just about three weeks after its predecessor 3.7 Flash and, according to Googles model card, is built on Gemini 3.7 Flash rather than a new base model. The multimodal model processes text, image, audio, video and PDF, keeps the context window of roughly 1M tokens (1,048,576) and delivers up to 64K output tokens. It is positioned for coding, agentic workflows and critical, multi-step reasoning. A hardened sibling variant called 3.8 Flash Cyber shipped alongside it.

Example / practical relevance

The introductory price is 0.75 USD per 1M input tokens and 3.75 USD per 1M output tokens — identical to 3.7 Flash; both rates double on 1 January 2027. In benchmarks the model beats 3.7 Flash and reaches, among others, 54.9% on HLE-Verified (multi-step reasoning) as well as leading scores on DeepSWE v1.1 (software engineering), Vals Finance Agent V2 and Harveys Legal Agent Benchmark. A characteristic behaviour: on complex tasks it executes extra reasoning steps and calls tools iteratively, consuming more tokens in the process. It is available in the Gemini app (Google AI Pro and Ultra), in AI Mode, in Google Sheets, and for developers in Google Antigravity, Google AI Studio and via the Gemini API.

Differentiation

Flash is Googles fast, cost-efficient line below the more powerful Pro models (such as Gemini 3 Pro): Pro targets maximum reasoning depth, Flash targets throughput and low cost per token. Compared with the direct predecessor 3.7 Flash, 3.8 Flash brings gains in coding, agents and multi-step reasoning at the same price level. The knowledge cutoff remains partly at March 2026 (earlier in other domains), which matters for questions about very recent events.

More on this topic

See everything in one place:Gemini Flash→

Encyclopedia · 12 Sept 2026Gemini Flash — Google's fast model line | LexiconGemini Flash is Google's fast, low-cost model line: high throughput and native multimodality. Positioning, use profile and a selection guide.GlossaryGemini 3.7 FlashGemini 3.7 Flash (August 2026) is Googles fast workhorse model for coding and agents — 1M-token context, introductory price of 0.75/3.75 USD per million tokens, half the cost of 3.6 Flash.GlossaryGemini 3 FlashGemini 3 Flash is Google's fast, low-cost thinking model from December 17, 2025 — near-Pro reasoning with a 1M-token context, built for agentic workflows and coding.