Gemini 3.5 Flash
Best for frontier performance across agents and coding
Solve complex real world problems at speed.
Intelligence in a Flash
Get advanced reasoning at Flash-level latency and scale.
Fast and smart
Speed and scale don’t have to come at the cost of intelligence.
Master complexity
Deep reasoning across long horizons and iterative coding tasks.
Truly multimodal
Frontier-level understanding across text, audio, images, code, and video.
Hands-on
Speed doesn’t come at the cost of intelligence. Here are just a few ways you can use Gemini 3.5 Flash’s agentic capabilities at speed.
Slide 1 of 9
Code faster through iterative loops
See how Gemini 3.5 Flash generates six payment UI options in under 60 seconds.
Develop multiple creative concepts in parallel
See how Gemini 3.5 Flash can create 64 fractal variations at a high speed.
Long horizon agentic execution
See how Gemini 3.5 Flash ingests the AlphaGo paper and builds an intelligent game autonomously.
Accelerate the creation of brand assets
Watch how Gemini 3.5 Flash coordinates multiple workflows to generate and refine a brand for a fundraiser with minimal input.
Generate interactive web animations in a single shot
See how Gemini 3.5 Flash turns a text description into fully interactive HTML components.
Create music through real-time collaboration
See how Gemini 3.5 Flash coordinates multiple agents to create a song using the Strudel music library.
Orchestrate multi-agent workflows
Watch Gemini 3.5 Flash coordinate a team of specialized agents to design and build a virtual city.
Organize large file collections efficiently
See how Gemini 3.5 Flash deploys parallel agents to automatically rename and structure messy datasets.
Build and improve a game with interactive agent loops
Watch Gemini 3.5 Flash deploy agents to continuously refine a game in real time.
Showcase
Slide 1 of 3
“Gemini's newest flash model advances the frontier for intelligence per dollar. It performs 42% better than Flash 3 on our long range, multi-turn cyber benchmark, while achieving a 68% improvement in token efficiency. This makes it an excellent choice for defenders seeking to scale horizontally. We are excited to continue collaborating with the Gemini team to help give Armadin customers the edge.”
David Slater, Chief Architect, Armadin
“Gemini 3.5 Flash beat Gemini 3 Flash by 19.6% on Box’s enterprise work evaluation set, which we designed to reflect the kinds of real-world, multi-step tasks our customers perform daily with the Box Agent. We’re already seeing stronger performance in workflows that matter to specific industries. Our Life Sciences customers can use Gemini 3.5 Flash to extract data and make calculations with 96.4% greater accuracy, helping them process scientific information more reliably. For Financial Services firms, where exact measurement drives trust, it can build financial reports from structured data with 46.7% greater accuracy.”
Ben Kus, Chief Technology Officer, Box
“Gemini 3.5 Flash delivers coding and reasoning quality close to Gemini Pro, while preserving the speed and cost profile that make Flash ideal for real-time developer workflows. The new model also improves low reasoning coding performance by 10–20% compared to the previous Flash generation.”
Nick Frolov, Head of Product, Junie, JetBrains
Performance
Our most impressive model yet for agentic workflows. Gemini 3.5 is leading across a wide range of benchmarks.
| Benchmark | Gemini 3.5 Flash | Gemini 3 Flash | Gemini 3.1 Pro | Claude Sonnet 4.6 | Claude Opus 4.7 | GPT-5.5 | ||
|---|---|---|---|---|---|---|---|---|
| Coding | Terminal-bench 2.1 Agentic terminal coding | Terminus-2 harness | 76.2% | 58.0% | 70.3% | — | 66.1% | 78.2% |
| SWE-Bench Pro (Public) Diverse agentic coding tasks | Single attempt | 55.1% | 49.6% | 54.2% | — | 64.3% | 58.6% | |
| Agentic | MCP Atlas Multi-step workflows using MCP | 83.6% | 62.0% | 78.2% | 69.5% | 79.1% | 75.3% | |
| Toolathlon Real-world general tool use | 56.5% | 49.4% | — | — | — | 55.6% | ||
| UI Control | OSWorld-Verified Agentic computer use | 78.4% | 65.1% | 76.2% | 72.5% | 78.0% | 78.7% | |
| Expert tasks | Finance Agent v2 Financial analysis and decision-making | 57.9% | 42.6% | 43.0% | 51.0% | 51.5% | 51.8% | |
| GDPval-AA Economically valuable knowledge work | Elo | 1656 | 1204 | 1314 | 1676 | 1753 | 1769 | |
| Multimodal | CharXiv Reasoning Information synthesis from complex charts | No tools | 84.2% | 80.3% | 83.3% | 72.4% | 82.1% | 84.1% |
| MMMU-Pro Multimodal understanding and reasoning | No tools | 83.6% | 81.2% | 80.5% | 74.5% | 75.2% | 81.2% | |
| Blueprint-Bench 2 Agentic spatial reasoning | Normalized score | 33.6% | 0.0% | 26.5% | 6.7% | 24.5% | 36.2% | |
| Long context | MRCR v2 (8-needle) Long context performance | 128k (average) | 77.3% | 67.2% | 84.9% | 84.9% | 59.3% | 94.8% |
| 1M (pointwise) | 26.6% | 22.1% | 26.3% | — | — | — | ||
| Reasoning | Humanity’s Last Exam Academic reasoning (full set, text + MM) | 40.2% | 33.7% | 44.4% | 33.2% | 46.9% | 41.4% | |
| ARC-AGI-2 Abstract reasoning puzzles | 72.1% | 33.6% | 77.1% | 58.3% | 75.8% | 84.6% |
For details on our evaluation methodology please see deepmind.google/models/evals-methodology/gemini-3-5-flash
Model information
- Name
- 3.5 Flash
- Status
- Preview
- Input
- Text
- Image
- Video
- Audio
- Output
- Text
- Input tokens
- 1M
- Output tokens
- 64k
- Knowledge cutoff
- January 2025
- Tool use
- Function calling
- Structured output
- Search as a tool
- Code execution
- Best for
- Everyday tasks
- Agentic coding
- Advanced reasoning
- Multimodal understanding
- Long context understanding
- Availability
- Gemini App
- Gemini API
- Gemini Enterprise
- Gemini Enterprise Agent Platform
- Google AI Mode
- Google AI Studio
- Google Antigravity
- Android Studio
- Documentation
- View developer docs
- Model card
- View model card