Google launches Gemini 3.5 Flash and new Omni multimodal AI family at I/O 2026
Google launched Gemini 3.5 Flash today as the default model for its Gemini app and AI Mode in Search, with Gemini 3.5 Pro following next month. The company also introduced Gemini Omni, a new multimodal AI family capable of generating video from text, photos, video, and audio inputs.
Gemini 3.5 Flash at a glance
How Gemini 3.5 Flash ranks now
| Benchmark | Score | Rank | Source |
|---|---|---|---|
| Arena Elo | 1478 Elo | 26 of 141 | LMArena, Nov 2026 ↗ |
| GPQA | 92.2% | 25 of 164 | Artificial Analysis, Jul 2026 ↗ |
| Speed (tok/s) | 167 tok/s | 17 of 103 | Artificial Analysis, Nov 2026 ↗ |
Rank is against every model in our database with a sourced score on that benchmark, each counted once with its best score.
Google launches Gemini 3.5 Flash and new Omni multimodal AI family at I/O 2026
Google released Gemini 3.5 Flash today at its I/O 2026 developer conference, making it the default model for the Gemini app and AI Mode in Search. Gemini 3.5 Pro will follow next month.
According to Google, Gemini 3.5 Flash is significantly faster than previous versions, offers improved agentic coding capabilities, and can generate "richer, more interactive web UIs and graphics." The company claims the model has improved guardrails to reduce harmful content generation and false positives on safe queries.
Gemini Omni: New multimodal family
Google introduced an entirely new AI model family called Gemini Omni. The first model, Omni Flash, is rolling out today in the Gemini app, Google Flow, and YouTube Shorts.
Unlike Google's existing Veo model (text-to-video only), Omni Flash can generate video clips from multiple input types: text, photos, video, and audio. Google says future versions will be able to "create anything from any input," though no timeline was provided.
Gemini Spark: Background AI agent
Google launched Gemini Spark, a continuously running AI agent powered by Gemini 3.5 Flash. The service runs 24/7 on virtual machines in Google Cloud and can connect to Google Workspace apps (Docs, Gmail, Sheets, Slides) and third-party services including Canva and Instacart.
Google plans to add access to local files through the Gemini app on macOS, though no release date was specified.
Android app generation in AI Studio
Google AI Studio now allows users to generate full native Android apps from prompts. The feature includes an embedded Android emulator for previewing apps, with options to export to Android Studio, GitHub, or ZIP files. Users can also test apps directly on connected phones.
Google says users will "soon" be able to publish AI-generated apps privately to friends and family. Firebase integration support is coming later.
Search and Gmail updates
Google is launching "Gmail Live," a voice-driven search interface for email that extracts specific information based on spoken queries. Similar features will extend to Google Docs and Keep.
Search is adding "information agents" that provide summarized updates on topics, plus "generative UI" that creates interactive visualizations and mini apps for repeated searches. These features launch this summer, starting with AI Pro and Ultra subscribers.
Pricing changes
Google cut AI Ultra subscription pricing from $249.99 to $100 per month, with a $200/month tier also available. The pricing change brings Google closer to OpenAI's pricing structure.
What this means
Google's launch of a dedicated Omni multimodal family signals intensifying competition in cross-modal AI generation, directly challenging OpenAI's GPT-4o and Anthropic's multimodal Claude models. The aggressive price cut for AI Ultra—dropping from $249.99 to $100/month—suggests Google is prioritizing market share over margins as enterprise AI adoption accelerates. The emphasis on agentic capabilities (Spark) and developer tools (AI Studio app generation) indicates Google is positioning for a future where AI acts autonomously rather than responding to individual prompts.
Related Articles
Google releases Genie 4 to AI Ultra subscribers with real-time prompt editing and native audio
Google is rolling out Genie 4, which it calls its most realistic and interactive world model yet, to eligible Google AI Ultra subscribers in Project Genie. The update adds real-time native audio, mid-session text prompting, a world-building agent and a community gallery. Genie 3 also gains sessions of up to 3 minutes.
Tencent releases Hy4 flash: 220B MoE, 12B active parameters, 1M-token context
Tencent has released Hy4 flash, a multimodal Mixture-of-Experts model with 220B total parameters, 12B active per token, and a 1,048,576-token context window. Tencent calls it its most cost-effective Hy model to date and claims a per-task cost of 1/20 that of Hy4 on its internal WorkIndex benchmark.
Tencent releases Hy4 Flash Base: 220B-parameter MoE, 12B active, 1M-token context
Tencent has published Hy4 Flash Base on Hugging Face: a 220B-parameter mixture-of-experts model with 12B active parameters per token, text and image input, and a 1M-token context. Tencent calls it its most cost-effective Hy model to date.
SenseNova-Loop-9B released: 9B parameters, 56 layer executions, 512K context
SenseNova has released SenseNova-Loop-9B, a roughly 9B-parameter multimodal model that runs 56 decoder-layer executions per forward pass from 32 distinct layers. It supports up to 512K tokens of context and, per the model card, needs about 15 GiB of GPU memory at 256K context with 4-bit weights and an 8-bit KV cache.
Comments
Loading...