Google just launched Gemini 3.5 Flash — here's all the upgrades
It looks like Google is preparing "Gemini Spark" to turn AI into an Operating System
www.tomsguide.comHere’s the latest on Gemini 3.5 Flash.
Short answer
- Gemini 3.5 Flash is Google’s fastest, low-latency model in the Gemini 3.5 family, and it’s being deployed as the default model in the Gemini app and AI Mode in Search. It emphasizes speed, efficiency, and strong coding/agent capabilities, with Gemini 3.5 Pro following for broader enterprise-grade features.
Details
- Release context and goals
- Google announced Gemini 3.5 as part of the Gemini 3.5 family, designed to combine frontier-level intelligence with action while prioritizing speed and cost efficiency. This positions 3.5 Flash as the default consumer-facing model in the Gemini app and AI Mode, with agentic features and improved token efficiency.[3][4]
- Key capabilities of Gemini 3.5 Flash
- Strong coding support and enhanced agent capabilities, with a focus on reducing latency and improving output throughput (claims of high tokens-per-second performance versus peers).[2][3]
- The model family includes a later 3.5 Pro variant aimed at more demanding enterprise use, expected to roll out publicly soon after the Flash release.[3]
- Availability and rollout
- Gemini 3.5 Flash is available to users through the Gemini app and Google AI services, with live deployment and broader tester access as part of the I/O 2026 announcements. Gemini Spark (an individual AI agent) uses 3.5 Flash in its operation.[3]
- Reports note 3.5 Flash is the default in AI Mode for Search and in the Gemini app, with ongoing expansion to testers and potential beta access for higher-tier subscribers.[5][3]
- Benchmarks and performance (contextual)
- Early benchmarks and third-party coverage describe 3.5 Flash as outperforming earlier Gemini 3.1 Pro on several tasks, and delivering markedly lower latency and faster response times relative to prior generations and competing models.[1][6][2]
- Some coverage highlights extremely favorable speed metrics (tokens per second) and cost-efficiency relative to larger Pro models, though exact figures vary by test and workload.[1][2][5]
- What’s next
- Google plans to roll out Gemini 3.5 Pro publicly in June, expanding capabilities and scaling for developers and enterprise users.[3]
- The Gemini API and developer tooling continue to evolve alongside 3.5, with release notes and changelogs documenting updates to models and APIs.[7][10]
Illustrative example
- If you’re using the Gemini app for everyday tasks, 3.5 Flash should feel noticeably faster in generating code snippets, composing multi-step actions, and handling typical AI Agent workflows, with Spark able to act on your directions in the background when enabled.[3]
Would you like a quick comparison table of Gemini 3.5 Flash vs 3.5 Pro vs Gemini 3 Pro across latency, cost, and typical use cases? I can also pull out the exact rollout dates and where to access Gemini 3.5 Flash in your region if you’d like.[6][3]