Models & Labs
Google Unveils Gemini 3.5 Flash Model
Matt Wolfehigh confidence
Why it matters
- →Provides a faster, cost-effective AI model option.
- →Enhances Google's competitive edge in AI model offerings.
- →Sets the stage for future advancements with the upcoming Pro version.
At Google I/O, the company announced the release of Gemini 3.5 Flash, the first model in the new Gemini 3.5 family. This model is designed to offer faster performance and cost efficiency. It is currently available, with a more advanced version, Gemini 3.5 Pro, expected to launch later. This release marks a significant step in Google's AI model development.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1Gemini 3.1 Pro Model Released
AI Explained · February 20, 2026 · Same story
- 2Google Launches Gemini 3.1 and Flash-Lite
Lev Selector · March 6, 2026 · Same story
- 3Google Unveils Gemini Deep Research Max
Matt Wolfe · April 24, 2026 · Related
- 4Google Enhances Gemini App with New AI Features
TechCrunch AI · May 19, 2026 · Same story
- 5Google I/O Showcases Gemini's Agentic Integration
The Rundown AI · May 20, 2026 · Related
- 6Google I/O 2026: Gemini 3.5 and Omni Unveiled
Google AI Blog · May 20, 2026 · Same story
- 7Gemini 3.5 Flash Prioritizes Speed Over Cost
The AI Daily Brief · May 20, 2026 · Related
- 8Google I/O 2026 Unveils Gemini Omni and More
Lev Selector · May 22, 2026 · Related
What happened next
- 9Google Unveils Gemini Omni and Gemini 3.5 Models
Google AI Blog · May 29, 2026 · Same story
- 10Google's Gemini 3.5 Pro Model Still Missing
The Rundown AI · July 22, 2026 · Same story
- 11Google Introduces Gemini Flash Models
Matt Wolfe · July 24, 2026 · Same story
- 12Google Launches Gemini 3.8 Flash Model
The Verge AI · September 2, 2026 · Same story
- 13Google Releases Gemini 3.8 Flash Model
Matthew Berman · September 3, 2026 · Same story
Follow this story
Open the full story →Google DeepMind Launches Gemini 3.5 Flash Model
17 developments
- May 15 · Google DeepMind
Google DeepMind Launches Gemini 3.5 Flash Model - May 17 · Google DeepMind
Google DeepMind Launches Gemini for Science↳ Gemini for Science introduces Hypothesis Generation, Computational Discovery, and Literature Insights tools to streamline scientific researc - May 17 · Google DeepMind
Google DeepMind Launches Gemini Omni for Video Creation↳ Google DeepMind launches Gemini Omni for video creation, allowing users to generate and edit videos from mixed inputs via natural language. - May 18 · The AI Daily Brief
Google I/O to Feature Gemini AI Models↳ Google I/O will unveil Gemini Spark and cost-optimized Gemini Flash models. - May 19 · Google AI Blog
Google Unveils Gemini Omni and 3.5 Models - May 19 · Google AI Blog
Google Unveils Gemini 3.5 with Enhanced AI Capabilities - May 19 · WIRED AI
Google Introduces Avatar Feature for AI Video Creation↳ Google Flow adds digital avatar creation powered by Omni Flash, using scans of likeness and voice to produce lifelike videos without physica - May 19 · TechCrunch AI
Google Unveils Gemini Omni for Multimodal Content Creation - May 19 · Google AI Blog
Google AI Launches New Subscription Plans at I/O 2026↳ Google unveils $100/month AI Ultra plan at I/O 2026 offering 5X usage limits, Gemini Spark agent, and 20TB storage. - May 19 · TechCrunch AI
Google Launches Gemini 3.5 Flash for AI Agents - May 19 · The Verge AI
Google I/O 2026: Gemini 3.5 and Omni Models Unveiled - May 19 · GitHub Changelog
Gemini 3.5 Flash Now Available on GitHub Copilot↳ GitHub integrates Gemini 3.5 Flash into Copilot for Pro, Business, and Enterprise users across VS Code and JetBrains IDEs. - May 19 · WIRED AI
Google I/O 2026: AI Agents and Gemini Updates - May 20 · AI Explained
Google Unveils Gemini 3.5 Flash Model - May 21 · WIRED AI
Google's Gemini App Introduces AI Avatar Tool↳ Google's Gemini app introduces an AI avatar tool using the Omni video model to create digital user likenesses for video content. - May 22 · Matt Wolfe
Google Unveils Gemini 3.5 AI Model - May 23 · Matt Wolfe
Google Unveils Gemini 3.5 Flash Model (This article)
More in Models & Labs
Models & Labsother
llama.cpp b12340 adds CUDA 13 and ROCm 10.0
This release quietly closes the hardware gap for local inference by adding default support for CUDA 13 and ROCm 10.0 across Linux and Windows. Users with newer NVIDIA GPUs or AMD accelerators no longer need to compile from source to get performance, while Apple Silicon builds now ship KleidiAI disabled by default, shifting the focus back to standard ARM optimizations. The inclusion of Snapdragon support for Linux and Android marks a significant step toward on-device AI beyond the x86/ARM desktop ecosystem. It is less about new features and more about ensuring llama.cpp remains the universal runtime for every available accelerator.
llama.cpp Releases
Models & Labscoding
llama.cpp b12342 adds few-row MMQ for speculative decoding
Speculative decoding just got a significant speed boost on NVIDIA GPUs. This release introduces a specialized kernel for 'few-row' matrix multiplications, targeting the small batch sizes typical of verification passes in speculative decoding workflows. By optimizing these specific operations for quantized formats like Q8_0 and IQ4 variants, llama.cpp reduces latency where it matters most during token generation. The change is CUDA-only for now but leverages hardware capabilities from Ampere onwards, making local inference faster without requiring new model architectures or complex configuration changes.
llama.cpp Releases
© TechCrunch AIModels & Labsmodels
Google deploys Gemini 4 Hydrogen for faster Search
Google is quietly shifting its search infrastructure by replacing older fast models with Gemini 4 Hydrogen, a tier designed specifically for speed and cost-efficiency rather than raw reasoning power. This model completes tasks 2.5 times faster than its predecessor while beating the more powerful Argon on specific intelligence benchmarks, proving that specialized efficiency can outperform brute force in high-volume workflows. By integrating this into Search's fan-out technique, Google now breaks down complex queries into subqueries with greater precision and speed, delivering richer visual responses without the latency of heavier models. This marks a strategic pivot toward agentic execution layers within consumer search, making AI Mode feel snappier for everyday multi-part questions.
TechCrunch AI