Skip to content
TechPP

Festive sale: the best deals in one sheet. See the deals

HomeUpdate TrackerGemini Updates

Gemini Update Tracker

Being a leader in the technology space, it's not surprising that Google is also in the AI race. However, it is slightly behind the others, including Claude and OpenAI. Google first introduced its model called Bard and then later changed its name to Gemini. Currently, Google also offers some of the best models for coding, image generation and much more.

Similar to OpenAI and Claude, Google also frequently releases model updates. The best part is Google AI models are bundled with the Google Gemini One Pro subscription, which also offers other benefits.

If you have been tracking Google Gemini model updates, you have come to the right place. In this post, you will find all the latest updates to Google Gemini models, the new announcements, and everything you need to know about the latest versions.

So as always, you can bookmark this post to get the latest Gemini updates.

Google Gemini Logo

Gemini Model Releases

Gemini 4 Argon

September 30, 2026
  • NEW
    Release:Google announced Gemini 4 Argon, its new frontier model and a significant upgrade over Gemini 3.1 Pro
  • UPDATE
    Availability:Rolling out in phases, starting with trusted cybersecurity defenders through Google's Fairwind Program before wider access
  • BENCHMARK
    Coding:Scores 77.9% on DeepSWE v1.1 and 57.4% on Terminal-Bench 4.0, up from 73.7% and 19.1% for Gemini 3.8 Flash
  • BENCHMARK
    Agents:Reaches 69.2% on OSWorld 2.0, up from 59.0%
  • CAPABILITY
    Output:Supports up to 1M output tokens in a single response, against 64K for Gemini 3.8 Flash
  • NEW
    Pricing:Introductory pricing of $2 per million input tokens and $10 per million output tokens

Gemini 3.8 Live

September 15, 2026
  • NEW
    Release:Google released Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, voice models that keep talking while they work
  • CAPABILITY
    Agents:Reasons in the background and calls tools asynchronously while audio keeps streaming, with spoken progress updates during multi-step tasks
  • BENCHMARK
    Audio:Google reports 82.6 on the Artificial Analysis Speech-to-Speech Quality Index, 68.6% on τ-Voice and 97.7% on Big Bench Audio for Extended Thinking
  • NEW
    Availability:Available in the Gemini API and Google AI Studio, and starting to roll out in Gemini Live

Lyria 3.5

September 3, 2026
  • NEW
    Release:Google introduced Lyria 3.5 as its next-generation music generation model
  • CAPABILITY
    Audio:Generates full-length songs with improved musical coherence and fine-grained control over duration and structure
  • CAPABILITY
    Multimodal:Takes text and image prompts and outputs high-fidelity 44.1 kHz stereo audio

Gemini 3.8 Flash

September 2, 2026
  • NEW
    Release:Gemini 3.8 Flash is generally available as Google's most intelligent Flash model, built for long-horizon software engineering, autonomous agents, and complex enterprise workflows
  • NEW
    Family:A locked-down security variant, Gemini 3.8 Flash Cyber, shipped alongside it
  • BENCHMARK
    Coding:Scores 73.7% on DeepSWE v1.1, up from 65.3% for Gemini 3.7 Flash
  • BENCHMARK
    Agents:Reaches 59.0% on OSWorld-2.0, up from 50.6%, and 56.5% on the hard tier of BioMysteryBench, up from 43.5%
  • CAPABILITY
    Context:Handles text, image, audio, video, and PDF input with a 1 million token context window and 64K output tokens
  • NEW
    Pricing:Introductory pricing of $0.75 per million input tokens and $3.75 per million output tokens until December 31, 2026, then $1.50 and $7.50 from January 1, 2027
  • NEW
    Availability:Live as gemini-3.8-flash in Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform

Agentic video understanding

September 1, 2026
  • NEW
    Video:Dynamic video timeline navigation rolled out for Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite, letting the model seek through a timeline instead of processing it statically
  • IMPROVED
    Efficiency:Google reports up to 88% fewer tokens on long-form video compared with static processing

Gemini Omni Flash (GA)

August 27, 2026
  • NEW
    Release:gemini-omni-1.1-flash reached general availability for video generation workflows
  • NEW
    Video:Adds video extension, interpolation between images, and configurable output resolutions from 360p to 4K
  • UPDATE
    Migration:The Omni Flash preview endpoint is scheduled for deprecation on September 30, 2026

Gemini 3.5 Transcribe

August 26, 2026
  • NEW
    Audio:Google introduced Gemini 3.5 Transcribe for live and prerecorded speech transcription
  • CAPABILITY
    Languages:Supports more than 85 languages with smart cleanup, speaker identification, and timestamps
  • NEW
    Availability:Available through Google's developer APIs for live and prerecorded audio workflows

Gemini 3.7 Flash

August 13, 2026
  • NEW
    Release:Google made Gemini 3.7 Flash generally available as its workhorse model for coding and agentic tasks
  • IMPROVED
    Coding:Scores 65.3% on DeepSWE v1.1 on Google's own evaluations, up from 49.0% for Gemini 3.6 Flash
  • CAPABILITY
    Context:Ships with a 1,048,576 token context window
  • NEW
    Pricing:Introductory pricing of $0.75 per million input tokens and $3.75 per million output tokens until December 31, 2026
  • NEW
    Availability:Live through the Gemini API in Google AI Studio, Android Studio, Google Antigravity, and the Gemini Enterprise Agent Platform and app

Gemini 3.6 Flash

July 21, 2026
  • NEW
    Release:Google released Gemini 3.6 Flash with stronger coding, knowledge, and multimodal performance
  • IMPROVED
    Efficiency:Google reports 17% fewer output tokens than Gemini 3.5 Flash for comparable tasks
  • NEW
    Family:The release was accompanied by Gemini 3.5 Flash-Lite and Gemini 3.5 Flash Cyber

Gemini 3.5 Flash GA & new models (late June 2026)

June 26, 2026
  • UPDATE
    Release:Gemini 3.5 Flash reaches general availability as Google's most intelligent model for sustained frontier performance on agentic and coding tasks
  • NEW
    Agents:Gemini 3.5 Flash adds Computer Use, letting you build custom agents that see, reason, and take action across desktop, mobile, and browser environments
  • NEW
    Audio:Gemini 3.5 Live Translate is a new speech-to-speech model that auto-detects 70+ languages while preserving the speaker's natural intonation
  • NEW
    Media:Nano Banana 2 Lite launches as the fastest, most cost-efficient Gemini image model, and Gemini Omni Flash arrives in API public preview for custom video workflows
  • NEW
    Local:Gemma 4 12B, Google's latest open model, runs locally in about 16GB of memory with a unified architecture that includes vision and native voice
  • NEW
    Chrome:Gemini in Chrome on Android (with Nano Banana and auto browse) rolls out to devices with 4GB of RAM or more, starting in US English

Gemini ecosystem updates (June 2026)

June 10, 2026
  • UPDATE
    Enterprise:Gemini 3.5 Flash is now enabled by default for all Gemini Enterprise users and cannot be disabled; Gemini 3.1 Pro and 3 Flash are in Limited Availability
  • FEATURE
    Reasoning:Wider rollout of Gemini 3 Deep Think for stronger reasoning, plus Deep Research, Gemini Live, and Gemini Omni across Android, Search, and YouTube
  • UPDATE
    API:The gemini-3.1-flash-image-preview and gemini-3-pro-image-preview models are deprecated and shut down on June 25, 2026

Gemini 3.5 Flash

May 19, 2026
  • NEW
    Release:Launched at Google I/O 2026 as Google’s new primary fast-response model, pairing Pro-level reasoning with Flash-class inference speed
  • CAPABILITY
    Multimodal:Native multimodal input across text, images, audio, and video with near real-time conversational performance
  • IMPROVED
    Performance:Produces output at roughly 4x the token rate of some frontier models, built for high-volume, low-latency workloads
  • BENCHMARK
    Reasoning:Scores 90.4% on GPQA Diamond (PhD-level science), 81.2% on MMMU-Pro (multimodal), and 78% on SWE-bench Verified (coding)
  • BENCHMARK
    Agents:Reaches 76.2% on Terminal-Bench 2.1, 83.6% on MCP Atlas, and 1656 Elo on GDPval-AA, outperforming Gemini 3.1 Pro on several metrics
  • NEW
    Availability:Rolling out across the Gemini app, AI Mode in Search, Workspace, and Android; available to developers via Google AI Studio, Vertex AI, the Gemini API, and Android Studio
  • UPDATE
    Roadmap:Gemini 3.5 Pro held back for internal use at launch, with a public rollout expected in June 2026

Gemini 3.1 Flash-Lite (GA)

May 7, 2026
  • NEW
    Release:Gemini 3.1 Flash-Lite reaches General Availability, the fastest and most cost-efficient model in the Gemini 3.x family
  • IMPROVED
    Performance:Optimized for ultra-low latency, sub-second to ~1.8s p95, and high-volume production workloads
  • PRICING
    Cost:Approximately 60% lower cost compared to thinking-tier models; matches Gemini 2.5 Flash on key capability benchmarks
  • UPDATE
    Migration:The older gemini-3.1-flash-lite-preview model deprecated simultaneously, with shutdown scheduled May 25, 2026

Lyria 3 Pro

Apr 24, 2026
  • NEW
    Release:New AI music generation model released as part of the April 2026 Gemini Drop
  • FEATURE
    Audio:Generates up to 3-minute music tracks with high-fidelity sound mixing; available in the Gemini app
  • NEW
    Availability:Available directly in the Gemini app for all users

Deep Research Max + Gemini Embedding 2

Apr 21, 2026
  • NEW
    Research:Deep Research Max launched as a next-generation autonomous research agent via the Interactions API; built on Gemini 3.1 Pro with Model Context Protocol (MCP) support
  • FEATURE
    Research:Deep Research Max adds native charts, infographics, and multimodal grounding; uses extended test-time compute to iteratively reason, search, and refine answers
  • NEW
    Embeddings:Gemini Embedding 2 reached General Availability, production-ready multimodal embeddings supporting text, image, video, and audio for search and reasoning tasks

Gemini 3.1 Flash TTS Preview

Apr 13, 2026
  • NEW
    Release:New text-to-speech model in the Gemini 3.1 family
  • FEATURE
    Capabilities:Built for controllable speech generation
  • UPDATE
    Migration:Replaces older 2.5 preview TTS direction in the latest Gemini API docs

Gemini 3.1 Flash Live Preview

Mar 11, 2026
  • NEW
    Multimodal:New live multimodal model path for real-time interactions
  • UPDATE
    Migration:Replaces older 2.0 and 2.5 live model paths in current docs
  • IMPROVED
    Focus:Built for low-latency live voice and interactive use cases

Gemini 3.1 Flash-Lite Preview

Mar 3, 2026
  • NEW
    Release:Latest Flash-Lite direction in the Gemini API docs
  • IMPROVED
    Performance:Built as the cheap and very fast 3.1-series model
  • UPDATE
    Migration:Designed to replace earlier 2.5 Flash-Lite preview paths

Gemini 3.1 Pro

Feb 19, 2026
  • NEW
    Positioning:Current top Gemini reasoning model in public materials
  • IMPROVED
    Capabilities:Built for broad world knowledge and advanced reasoning across modalities
  • IMPROVED
    Agents:Better tool use and stronger multi-step agentic work than earlier Gemini 3 models
  • BENCHMARK
    Reasoning:Reached 44.4% on Humanity’s Last Exam in one Google benchmark table
  • BENCHMARK
    Reasoning:Reached 77.1% on ARC-AGI-2
  • BENCHMARK
    Knowledge:Reached 94.3% on GPQA Diamond
  • NEW
    Release:Positioned as Google’s most intelligent Gemini model at the time of release

Gemini 3 Pro

Late 2025
  • NEW
    Positioning:Top-end Gemini 3 generation model
  • IMPROVED
    Performance:Improved depth, reasoning, and reliability over Gemini 2.5 Pro
  • BENCHMARK
    Coding:Google says it solves more than 50% more benchmark tasks than Gemini 2.5 Pro in developer-tool scenarios
  • NEW
    Focus:Built for heavier coding and advanced reasoning work

Gemini 3 Flash

Late 2025
  • NEW
    Release:First 3-series Flash model
  • IMPROVED
    Performance:Introduced as a Flash model with Pro-level intelligence at Flash speed and pricing
  • NEW
    Focus:Built to push the fast model tier much closer to the top-end tier
  • NEW
    Availability:Became the latest 3-series fast direction in Gemini developer docs

Gemini 2.5 family overall

Late 2025
  • IMPROVED
    Focus:Thinking became a major part of the Gemini line
  • IMPROVED
    Performance:Stronger coding and advanced reasoning than Gemini 2.0
  • FEATURE
    Multimodal:Broader multimodal support across text, image, audio, video, and documents
  • NEW
    Capabilities:Computer-use and native audio directions were added to the family roadmap
  • UPDATE
    Availability:Gemini 2.5 Pro and Flash became the main public Gemini API models for advanced and fast use cases

Gemini 2.5 Flash Image

Oct 2, 2025
  • NEW
    Release:Added a Flash-class model focused on image generation and editing
  • FEATURE
    Image Generation:Brings native image generation into the 2.5 Flash family
  • IMPROVED
    Cost:Built for faster and cheaper image workloads than higher-end image models

Gemini 2.5 Flash-Lite

Jul 22, 2025
  • NEW
    Positioning:Fastest and most budget-friendly model in the Gemini 2.5 family
  • IMPROVED
    Focus:Built for maximum speed and lowest cost
  • CAPABILITY
    Use Cases:Designed for classification, extraction, routing, and lightweight multimodal tasks
  • NEW
    Focus:Sits below 2.5 Flash for cost-sensitive production use

Gemini 2.5 Flash

Jun 17, 2025
  • NEW
    Release:First stable Gemini 2.5 Flash release
  • IMPROVED
    Performance:Best Gemini model in terms of price-performance in the 2.5 family
  • NEW
    Focus:Built for low-latency, large-scale, high-volume tasks that still need reasoning
  • CAPABILITY
    Agents:Supports agentic workflows better than earlier Flash models
  • UPDATE
    Positioning:Became the main fast model in the 2.5 line

Gemini 2.5 Pro update

May 20, 2025
  • IMPROVED
    Performance:Gemini 2.5 Pro got stronger after launch
  • FEATURE
    Reasoning:Added Deep Think, an experimental enhanced reasoning mode for harder math and coding tasks
  • FEATURE
    Audio:Added new capabilities planned for native audio output
  • FEATURE
    Security:Added advanced security safeguards
  • NEW
    Computer Use:Added Project Mariner computer-use capabilities to the 2.5 family direction

Gemini 2.5 Pro

Mar 25, 2025
  • NEW
    Release:First Gemini 2.5 model release
  • NEW
    Positioning:Introduced as Google’s most intelligent AI model at launch
  • FEATURE
    Reasoning:Built as a thinking model that reasons through its thoughts before responding
  • BENCHMARK
    Performance:Debuted at #1 on LMArena in Google’s launch materials
  • NEW
    Positioning:Positioned as the top Gemini model for reasoning, coding, and complex multimodal work

Gemini 2.0 Flash general rollout

Jan 2025
  • UPDATE
    Release:Gemini 2.0 Flash moved from experimental availability toward wider production use
  • NEW
    Positioning:Became the new fast Gemini direction for chat, learning, brainstorming, and everyday assistant tasks

Gemini 2.0 Flash

Dec 11, 2024
  • NEW
    Release:First Gemini 2.0 model release
  • NEW
    Positioning:Built for the agentic era
  • FEATURE
    Multimodal:Supports multimodal input with text output at launch
  • FEATURE
    Generation:Native image generation and text-to-speech were introduced for early-access partners
  • IMPROVED
    Focus:Designed for lower latency and stronger tool-driven workflows than the 1.5 line

Gemini 1.5 Pro pricing and production update

Sep 24, 2024
  • UPDATE
    Release:Production-ready Gemini 1.5 models were updated again
  • PRICING
    Cost:Gemini 1.5 Pro input pricing was cut by 64% for prompts under 128K tokens
  • PRICING
    Cost:Output pricing was cut by 52%
  • PRICING
    Cost:Incremental cached token pricing was also cut by 64%
  • IMPROVED
    Usage:Rate limits were increased for developers

Gemini 1.5 Flash

May 14, 2024
  • NEW
    Release:First Flash model in the Gemini line
  • IMPROVED
    Focus:Built for lower latency and lower cost than 1.5 Pro
  • CAPABILITY
    Use Cases:Optimized for summarization, chat apps, captioning, data extraction, and fast large-scale tasks
  • IMPROVED
    Balancing:Keeps multimodal support while focusing more on speed and efficiency

Updated Gemini 1.5 Pro

May 14, 2024
  • IMPROVED
    Update:Improved over the earlier 1.5 Pro preview
  • IMPROVED
    Performance:Better at translation, coding, reasoning, and long-context tasks
  • UPDATE
    Availability:Expanded availability for developers
  • NEW
    Positioning:Positioned as the stronger long-context model in the 1.5 family

Gemini 1.5 Pro

Feb 15, 2024
  • NEW
    Release:First Gemini 1.5 model
  • NEW
    Architecture:Uses a Mixture-of-Experts architecture for better efficiency
  • IMPROVED
    Performance:Delivers performance similar to Gemini 1.0 Ultra in a smaller model class
  • FEATURE
    Context:Introduced a 1 million token context window for early testing
  • CAPABILITY
    Use Cases:Built for long-context work across text, image, audio, video, and code
  • CAPABILITY
    Multimodal:Can handle very large files and long multimodal prompts in one run

Gemini 1.0 Ultra / Pro / Nano

Dec 6, 2023
  • NEW
    Release:First Gemini model family release
  • NEW
    Positioning:Launched in three sizes: Ultra, Pro, and Nano
  • NEW
    Architecture:Built as a natively multimodal model from the start
  • IMPROVED
    Performance:Better at text, code, image, audio, and video understanding than Google’s earlier model line
  • NEW
    Positioning:Gemini Ultra became the top-end model in the first generation

Was this article helpful?

Comments

0

Share your thoughts, questions, or feedback about this guide.

Leave a Comment

0/2000