Mungomash LLC
Gemini Versions

2023 – 2026

Gemini Versions

Every Google Gemini release — from Bard's March 2023 launch and Gemini 1.0 in December 2023 through Gemini 3.1 Pro in February 2026 — with API model strings, ship dates, and the major changes per version. Plus the Bard–to–Gemini rename, the Brain × DeepMind merger, the February 2024 image-generation incident, the DOJ search-monopoly remedies that explicitly cover the Gemini app, the $40B Anthropic investment, and the Project Astra / Live / Mariner agentic surfaces.

Family & status

Family

Pro — flagship reasoning models, including the original Gemini 1.0 Ultra tier
Flash — speed- and cost-optimized chat models, including Flash-Lite and Flash-8B
Specialized — on-device (Nano) and dedicated image / multimodal output models
Pre-Gemini — PaLM 2 and the original LaMDA-based Bard, before the December 2023 Gemini rebrand

Status

Current — actively recommended; the latest in its family
Available — still served via API but superseded
Legacy — deprecated or sunset; no longer the recommended surface

Gemini version table

Model
Gemini 3.1 Pro
gemini-3.1-pro-preview
Pro
Current
Feb 19, 2026
Reasoning + agentic refresh of Gemini 3 Pro. ARC-AGI-2 jumped from 31.1% to 77.1%. 1M input / 65K output. Current Google flagship.
  • Released February 19, 2026; the announcement is at blog.google. Replaced gemini-3-pro-preview, which was deprecated on March 26, 2026.
  • ARC-AGI-2 jumped from 31.1% to 77.1% over the 3 Pro baseline; agentic-benchmark gains in the 45–80% range across coding, browser, and tool-use evals (vendor-stated).
  • 1,048,576-token input context; 65,536-token output cap. Multimodal across text, audio, images, video, and full code repositories.
  • Available across the Gemini app (AI Pro and AI Ultra tiers), NotebookLM, Google AI Studio, the Antigravity agent platform, Vertex AI, Gemini Enterprise, the Gemini CLI, and Android Studio at launch.
  • The model card is published on the DeepMind site.
Model
Gemini 3 Flash
gemini-3-flash-preview
Flash
Current
Dec 17, 2025
Default model in the Gemini app and Search AI Mode. “Pro-grade reasoning at Flash speed.” $0.50 / $3 per 1M tokens.
  • Released December 17, 2025; the announcement is at blog.google/products/gemini/gemini-3-flash.
  • Replaced Gemini 2.5 Flash as the default in the Gemini app and in Search's AI Mode.
  • Pricing $0.50 / $3.00 per million input / output tokens at launch.
  • Available simultaneously across the Gemini API, AI Studio, Antigravity, the Gemini CLI, Android Studio, Vertex AI, and Gemini Enterprise.
  • Marketed as “Pro-grade reasoning at Flash-level speed”; the smaller, cheaper companion to Gemini 3 Pro and (later) 3.1 Pro.
Model
Gemini 3 Pro
gemini-3-pro-preview — deprecated 2026-03-26
Pro
Legacy
Nov 18, 2025
First Google model launched simultaneously into Search, Gemini app, AI Studio, Vertex AI, and the Gemini CLI on day one. Top of LMArena at 1501 Elo at launch.
  • Released November 18, 2025; the announcement is at blog.google/products/gemini/gemini-3.
  • Day-one rollout across Search, Gemini app, AI Studio, Vertex AI, and the Gemini CLI — Google's first model launched simultaneously into Search.
  • Topped the LMArena Text Arena at 1501 Elo at launch; 37.5% on Humanity's Last Exam (no tools); 91.9% GPQA Diamond.
  • Introduced the Gemini 3 Deep Think mode, gated to AI Ultra subscribers and safety-tester cohorts. The mode generalized the parallel-thinking pattern from Gemini 2.5 Deep Think.
  • Superseded by Gemini 3.1 Pro on February 19, 2026; gemini-3-pro-preview formally deprecated and redirected to gemini-3.1-pro-preview on March 26, 2026.
Model
Gemini 2.5 Flash Image (“Nano Banana”)
gemini-2.5-flash-image
Specialized
Available
Aug 26, 2025
Image generation and natural-language editing inside the Gemini API. Character consistency, scene blending, local edits. SynthID watermarked.
  • Released August 26, 2025; the announcement is at developers.googleblog.com. Documented at ai.google.dev.
  • Character consistency, scene blending, and natural-language local edits — remove people, change pose, colorize, restyle — without the round-trip to a dedicated image-generation app.
  • Codenamed Nano Banana in pre-release testing; the codename stuck in much of the developer-community framing of the launch.
  • All output watermarked with SynthID; pricing approximately $30 per million output tokens (~$0.039 / image).
  • Distinct from the Imagen model line; covered briefly in the Imagen / Veo prose section below.
Model
Gemini 2.5 Deep Think
mode toggle within gemini-2.5-pro — AI Ultra only
Pro
Available
Aug 1, 2025
Parallel-thinking mode on top of Gemini 2.5 Pro. Generates and cross-evaluates many candidate solutions. AI Ultra subscribers only at $249.99/mo.
  • Released August 1, 2025; the announcement is at blog.google/products/gemini/gemini-2-5-deep-think. Model card at storage.googleapis.com (PDF).
  • Parallel thinking — the model generates many candidate ideas, considers them simultaneously, and revises or combines them before answering, rather than producing a single chain-of-thought trace.
  • Available only on the Google AI Ultra tier ($249.99/month) at launch, with a fixed daily prompt cap; first teased at Google I/O 2025.
  • State-of-the-art on LiveCodeBench V6 at launch; the same pattern was generalized into the gated Gemini 3 Deep Think mode on Gemini 3 Pro.
Model
Gemini 2.5 Flash-Lite
gemini-2.5-flash-lite
Flash
Available
Jul 22, 2025
Cheapest, fastest member of the 2.5 family. $0.10 / $0.40 per 1M tokens. 1M context. Routine recommendation for high-volume cheap-tier work.
  • Preview shipped June 17, 2025 alongside the 2.5 Pro / Flash GA; stable GA on July 22, 2025. The announcement is at developers.googleblog.com.
  • $0.10 / $0.40 per million input / output tokens at launch — the cheapest tier in the 2.5 family.
  • 1M-token context window; optimized for high-volume, low-latency workloads (classification, routing, summarization, agent steps).
  • Updated preview gemini-2.5-flash-lite-preview-09-2025 shipped September 25, 2025 with quality improvements; the stable id remained gemini-2.5-flash-lite.
Model
Gemini 2.5 Flash
gemini-2.5-flash
Flash
Available
Jun 17, 2025
The Flash sibling to 2.5 Pro. Default Gemini-app model from Google I/O 2025 until 3 Flash took over in December. Native audio output. Project Mariner computer use.
  • Preview unveiled at Google I/O 2025 on May 20, 2025; stable GA on June 17, 2025. The I/O announcement is at blog.google; thinking-mode updates at developers.googleblog.com.
  • Default Gemini-app model from I/O 2025 until Gemini 3 Flash took over the slot in December 2025.
  • 20–30% fewer tokens used in evals vs. the prior preview SKU; native audio output and “thought summaries” for enterprise auditability.
  • First Flash-tier model with Project Mariner computer-use capabilities folded in.
Model
Gemini 2.5 Pro
gemini-2.5-pro
Pro
Available
Mar 25, 2025
First Gemini family member to ship as a thinking model. Built-in chain-of-thought reasoning with a controllable thinking budget. 1M context.
  • Experimental released March 25, 2025; stable GA on June 17, 2025 after preview snapshots gemini-2.5-pro-preview-03-25, -05-06, and -06-05. The announcement is at blog.google.
  • First Gemini family member to ship reasoning as a default capability rather than a separate Thinking-Experimental SKU — built-in chain-of-thought, with a controllable thinkingBudget parameter.
  • 63.8% on SWE-Bench Verified at launch; 18.8% on Humanity's Last Exam (no tools).
  • 1M-token context window at launch; the 2M expansion that 1.5 Pro pioneered remained on Google's roadmap rather than shipping immediately.
  • The Deep Think mode on top of 2.5 Pro shipped four months later (August 1, 2025) on the AI Ultra tier; the page row above covers it.

The reasoning era — March 25, 2025. Above this line: the 2.5 and 3.x generations, where reasoning, multi-agent collaboration, native image generation, and agentic surfaces (Astra, Mariner, Antigravity) became first-class capabilities and the Gemini app moved to a one-launch-everywhere release pattern. Below: the 2.0 family's Thinking-Experimental separate-SKU era, and before that the 1.5 long-context era, the original 1.0 Ultra / Pro / Nano line, and the pre-Gemini PaLM 2 / Bard era.

Model
Gemini 2.0 Pro Experimental
gemini-2.0-pro-exp-02-05
Pro
Legacy
Feb 5, 2025
Largest context window of the era at 2M tokens. Top coding and world-knowledge model when shipped.
  • Released February 5, 2025 alongside 2.0 Flash GA and 2.0 Flash-Lite. The announcement is at blog.google.
  • 2,000,000-token context window — matched the 1.5 Pro 2M expansion of June 2024 and remained the largest Gemini context for several months.
  • Top of Google's coding and world-knowledge benchmarks for the 2.x family at launch; superseded by Gemini 2.5 Pro seven weeks later.
  • Shipped only as an Experimental SKU; never reached a stable gemini-2.0-pro id.
Model
Gemini 2.0 Flash + Flash-Lite
gemini-2.0-flash, gemini-2.0-flash-lite
Flash
Available
Feb 5, 2025
2.0 Flash GA. 2.0 Flash-Lite as the cost-optimized sibling. Outperformed 1.5 Pro at twice the speed.
  • Both stable ids shipped February 5, 2025; the developer-side announcement is at developers.googleblog.com.
  • Outperformed Gemini 1.5 Pro on most benchmarks at twice the speed by Google's accounting at launch.
  • 1M-token context window across both SKUs.
  • 2.0 Flash-Lite was positioned as the cost-optimized variant for high-volume work; later superseded by the 2.5 Flash-Lite family in July 2025.
Model
Gemini 2.0 Flash Thinking Experimental
gemini-2.0-flash-thinking-exp-1219
Flash
Legacy
Dec 19, 2024
Google's first reasoning model. Visible “thinking process” in the response. The OpenAI o1 response, three months after o1-preview.
  • Released December 19, 2024; competitive with OpenAI's o1 family that had launched September–December 2024.
  • Visible “thinking process” — the response surface exposed the model's chain-of-thought, the same pattern that Gemini 2.5 Pro later folded into the default Pro tier.
  • 32,767-token context at launch — substantially smaller than the 1M-token 2.0 Flash family it sat alongside.
  • Free in Google AI Studio at launch with a 2 RPM / 50 RPD limit; superseded by the integrated reasoning in Gemini 2.5 Pro three months later.
Model
Gemini 2.0 Flash (Experimental)
gemini-2.0-flash-exp
Flash
Legacy
Dec 11, 2024
Project Astra, Project Mariner, and Jules debut alongside. Native multimodal output (image, multilingual TTS) and Multimodal Live API.
  • Released December 11, 2024 in the “agentic era” framing post; the announcement is at blog.google.
  • Project Astra (universal AI agent prototype), Project Mariner (browser-controlling agent), and Jules (developer agent) all debuted alongside this release.
  • Native multimodal output: image generation and steerable, multilingual text-to-speech in a single API call.
  • Native tool use including Google Search and code execution; first model to expose the Multimodal Live API for real-time audio and video conversations.
  • Superseded by the GA gemini-2.0-flash on February 5, 2025.
Model
Gemini 1.5 Flash-8B
gemini-1.5-flash-8b
Flash
Legacy
Oct 3, 2024
Lowest cost-per-intelligence in the 1.5 family. 8B-parameter Flash sibling. 1M context.
  • Experimental released September 2024; stable GA on October 3, 2024. The announcement is at developers.googleblog.com.
  • Smallest Gemini SKU to ship a public API id at launch; positioned for the highest-volume cheap-tier agent steps.
  • 1M-token context window despite the smaller parameter count; superseded by the 2.0 / 2.5 Flash-Lite line.
Model
Gemini 1.5 — 2M context, “002” refresh
gemini-1.5-pro-002, gemini-1.5-flash-002
Pro
Legacy
Sep 24, 2024
2M context for 1.5 Pro (June 27, 2024). Stable “002” refresh of Pro and Flash with reduced 1.5 Pro pricing (September 24, 2024).
  • Two consecutive milestones rolled into one row. 2,000,000-token context for 1.5 Pro shipped June 27, 2024 alongside code execution; the announcement is at developers.googleblog.com.
  • “002” stable refresh of both 1.5 Pro and 1.5 Flash on September 24, 2024 with reduced 1.5 Pro pricing and increased rate limits; developer-blog post.
  • The 2M context expansion was the first time any production frontier model went beyond 1M; the milestone became the routine framing for “long-context” through 2024 and most of 2025.
Model
Gemini 1.5 Flash
gemini-1.5-flash, gemini-1.5-flash-001
Flash
Legacy
May 14, 2024
First Flash-tier release. 1M context. Multimodal text+image+audio+video input. Debuted at Google I/O 2024.
  • Released May 14, 2024 at Google I/O; the announcement is at blog.google.
  • First Flash-tier Gemini — the cheaper, faster sibling pattern that has been the routine recommendation for high-volume work in every Gemini generation since.
  • 1M-token context window; multimodal text + image + audio + video input on the same surface as 1.5 Pro.
  • Project Astra was first demoed at the same I/O; covered in the prose history section below.
Model
Gemini 1.5 Pro
gemini-1.5-pro, gemini-1.5-pro-001
Pro
Legacy
Feb 15, 2024
First Mixture-of-Experts in the family. 128K standard context, 1M in private preview. Equal or better to 1.0 Ultra at substantially less compute.
  • Announced February 15, 2024; the post is at blog.google. Public preview on Vertex AI followed April 9, 2024.
  • First Mixture-of-Experts model in the Gemini family — the architecture choice that made the 1M / 2M context expansions practical.
  • 128,000-token standard context at launch with 1,000,000-token context in private preview — the first production frontier model to ship a 1M-token surface.
  • Equal or better to Gemini 1.0 Ultra on most evaluations at substantially less inference compute, by Google's accounting.
Model
Gemini 1.0 Ultra (“Gemini Advanced”)
gemini-ultra — primarily product-only
Pro
Legacy
Feb 8, 2024
The largest and most capable Gemini 1.0 tier. Launch of Gemini Advanced subscription. Bard renamed to Gemini the same day.
  • Announced December 6, 2023 alongside the rest of the Gemini 1.0 family; general availability on February 8, 2024 via the new Gemini Advanced subscription on the $19.99/month Google One AI Premium plan. The announcement is at blog.google.
  • Reported 90.0% on MMLU at launch — Google's framing was that Ultra was the “first to outperform human experts” on the benchmark.
  • 32,768-token context window; primarily exposed through the consumer Gemini Advanced surface rather than as a routine API model.
  • Gemini Advanced was rebranded to Google AI Pro at I/O 2025 alongside the higher-priced AI Ultra tier ($249.99/month). The Bard–to–Gemini rename, also February 8, 2024, is covered in the prose history below.
Model
Gemini 1.0 Pro
gemini-pro, gemini-1.0-pro
Pro
Legacy
Dec 6, 2023
First public Gemini API model. Powered Bard from day one (170 countries, English). API GA December 13, 2023.
  • Announced December 6, 2023; the announcement is at blog.google/technology/ai/google-gemini-ai. Public API GA on December 13, 2023.
  • Powered Bard from day one in 170 countries (English at launch) — the model behind Bard for the two months between the December 6 launch and the February 8 Bard–to–Gemini rename.
  • 32,768-token context window; the first Gemini SKU developers could call from the public Gemini API.
  • Superseded by the Mixture-of-Experts 1.5 line two months later; deprecated through 2024.
Model
Gemini 1.0 Nano (and successors)
gemini-nano — on-device, AICore / ML Kit GenAI APIs
Specialized
Available
Dec 6, 2023
First Gemini on-device. Pixel 8 Pro at launch (Smart Reply, Summarize). Expanded across Pixel 8 / 8a / 9 / 10 and Chrome through 2024–2025.
  • Released December 6, 2023 on the Pixel 8 Pro Feature Drop — the first phone to run a frontier-lab model on-device. The post is at blog.google/products/pixel.
  • Initial features: Smart Reply in Gboard (WhatsApp at launch) and Summarize in Recorder; on Tensor G3 silicon.
  • Developer preview to Pixel 8 in April 2024; experimental access generally available across Android via AICore + ML Kit GenAI APIs in October 2024.
  • Expanded across Pixel 9 (multimodal Nano, August 2024), Pixel 8a, and Pixel 10 (August 2025) with successive on-device feature additions; later wired into Chrome's Built-in AI Prompt API.
  • This row consolidates the full on-device Nano lineage rather than splitting per-device-feature drops; treat it as the open-ended Specialized track within the Gemini family.

The Gemini era — December 6, 2023. Above this line: the natively-multimodal Gemini family, Google's response to ChatGPT and to the Brain × DeepMind merger of April 2023. Below: the Pre-Gemini era — PaLM 2 in May 2023 as Bard's underlying model, and the original LaMDA-based Bard launched in March 2023 as Google's reactive answer to ChatGPT.

Model
PaLM 2
chat-bison, text-bison, code-bison — deprecated
Pre-Gemini
Legacy
May 10, 2023
Replaced LaMDA as Bard's underlying model. Four sizes (Gecko / Otter / Bison / Unicorn). 100+ languages. Med-PaLM 2 and Sec-PaLM specialty variants.
  • Announced May 10, 2023 at Google I/O 2023; the Bard-related framing is at blog.google.
  • Replaced LaMDA as the model underlying Bard; supported 100+ natural languages and 20+ programming languages.
  • Four internal size codenames — Gecko, Otter, Bison, Unicorn — exposed via the PaLM API as chat-bison, text-bison, code-bison, etc.
  • Foundation for the Med-PaLM 2 medical and Sec-PaLM security specialty models.
  • Deprecated entirely once Gemini 1.0 shipped in December 2023; the PaLM API was sunset alongside.
Model
Bard (LaMDA-based)
never had a public API; product-only
Pre-Gemini
Legacy
Mar 21, 2023
Google's initial ChatGPT response. LaMDA-based at launch. Renamed to Gemini on February 8, 2024.
  • Released March 21, 2023 in early access to US and UK users; the post is at blog.google/innovation-and-ai/products/try-bard.
  • LaMDA-based at launch; framed as “a complement to Search,” not a replacement.
  • Followed Google's February 2023 ChatGPT-response announcement and the widely-reported Bard demo gaffe (the “James Webb Telescope first exoplanet image” error in the launch demo video).
  • Initial rollout blocked in the EU over GDPR concerns; came online there in July 2023.
  • Renamed Gemini on February 8, 2024 alongside the Gemini Advanced subscription launch — bard.google.com began redirecting to gemini.google.com that day.

Click any row to expand. Each row has a stable id for sharing — e.g. /ai/gemini/versions/#gemini-3-1-pro, #gemini-2-5-pro, #bard. The current model list is at ai.google.dev/gemini-api/docs/models; release notes at ai.google.dev/gemini-api/docs/changelog; Vertex AI roster at cloud.google.com/vertex-ai.

The Bard launch (March 2023) and the Bard–to–Gemini rename (February 2024)

Google's reactive answer to ChatGPT was announced on February 6, 2023 and shipped to early-access US and UK users on March 21, 2023 as Bard, an experimental conversational AI service backed by LaMDA. The launch demo's well-publicized error — the model misattributed the first exoplanet image to the James Webb Space Telescope — helped frame the early-2023 narrative that Google was caught flat-footed by OpenAI's November 2022 ChatGPT release. EU access was blocked initially over GDPR concerns and came online in July 2023.

At Google I/O on May 10, 2023, Bard's underlying model swapped from LaMDA to PaLM 2. Specialty variants — Med-PaLM 2 for medical question answering, Sec-PaLM for security — followed. Duet AI for Workspace shipped at $30/user/month on August 29, 2023 as the productivity-suite analog. The PaLM API was deprecated entirely once Gemini 1.0 shipped in December 2023.

The Bard–to–Gemini rename happened on February 8, 2024, the same day Gemini 1.0 Ultra reached general availability via the new Gemini Advanced subscription on the $19.99/month Google One AI Premium plan. bard.google.com began redirecting to gemini.google.com; Duet AI for Workspace was rebranded Gemini for Workspace; a new Android Gemini app shipped that day. The post is at blog.google. Gemini Advanced was itself rebranded Google AI Pro at I/O 2025 on May 20, 2025 alongside the higher-priced Google AI Ultra tier ($249.99/month).

The Brain × DeepMind merger (April 2023)

On April 20, 2023, Sundar Pichai announced the merger of Google Brain and DeepMind into a single unit, Google DeepMind, with Demis Hassabis (DeepMind cofounder, then CEO) named CEO of the combined entity. The announcement is at blog.google/technology/ai/april-ai-update; DeepMind's side is at deepmind.google/blog/announcing-google-deepmind.

Jeff Dean, previously the Google Brain lead, was named Chief Scientist of Google + Alphabet, reporting directly to Pichai. The combined unit's accomplishments cited in the announcement: AlphaGo, the original Transformer paper, word2vec, WaveNet, AlphaFold, seq2seq, deep reinforcement learning, TensorFlow, and JAX. The merger is the corporate-side throughline that explains why Gemini was framed in December 2023 as “natively multimodal trained from the ground up” — Google's first model assembled by the unified post-merger team.

Demis Hassabis and DeepMind's John Jumper were jointly awarded the 2024 Nobel Prize in Chemistry with David Baker for AlphaFold's protein-structure-prediction work. The DeepMind announcement is at deepmind.google. Hassabis remains CEO of Google DeepMind through April 2026.

The February 2024 Gemini image-generation incident

Within days of the Bard–to–Gemini rename, users posted historically inaccurate images Gemini's image generator produced — racially diverse Nazi soldiers, multi-ethnic US Founding Fathers, Black female popes — that drew sharp criticism. Google paused people-image generation in Gemini on February 22, 2024. The explainer post by then-SVP for Search Prabhakar Raghavan said directly: “we got it wrong.”

On February 28, 2024 Pichai's internal memo — reported in NPR — called the outputs “completely unacceptable” and committed to structural changes, updated guidelines, and red-teaming. People-image generation was restored on August 28, 2024 with Imagen 3, gradually rolled out starting with Gemini Advanced / Business / Enterprise; TechCrunch coverage.

The episode is recorded here factually because it bears on the Gemini-1.0 / Imagen-2 rollout window and on Google's subsequent moderation framing for image generation. This page does not litigate the underlying decisions or their handling.

The Anthropic investments

Google has been Anthropic's primary cloud and capital partner alongside Amazon. The October 27, 2023 $2 billion commitment ($500M upfront, $1.5B over time) was reported in CNBC, alongside Anthropic's $3B Google Cloud commitment over four years and adoption of TPU v5e for Claude inference.

On April 24, 2026 — four days before this writing — Google announced an additional commitment of up to $40 billion in Anthropic ($10B initial, up to $30B follow-on) at a reported $350B enterprise value. Coverage in TechCrunch, CNBC, and Bloomberg. The Anthropic relationship is the corporate-strategy counterweight to the in-house Gemini line; for the Anthropic side, see Claude Leadership.

Project Astra, Gemini Live, and Project Mariner

Project Astra — the “universal AI agent” prototype with the smart-glasses demo — was first unveiled at Google I/O on May 14, 2024 as a research preview. The DeepMind page is at deepmind.google/models/project-astra. Astra's multimodal real-time understanding capabilities folded into Gemini Live (consumer) and into Search's AI Mode (May 2025).

Gemini Live, the conversational voice surface (10 voices, interruptible, runs in background), launched alongside the Pixel 9 series on August 13, 2024, initially as a Gemini Advanced subscriber feature. Camera and screen-sharing capabilities followed in 2025 as Astra's surface migrated into the Gemini app.

Project Mariner — the Gemini 2.0–powered research-prototype Chrome extension that controls the browser (cursor, clicks, forms) — was announced December 11, 2024 alongside the Gemini 2.0 Flash Experimental release. Single-tab and foreground-only at launch; 83.5% on the WebVoyager benchmark. Mariner's computer-use capability was folded into Gemini 2.5 Flash by I/O 2025 and into the Antigravity agent platform that launched alongside Gemini 3 Pro.

Imagen and Veo

The image and video lines under the Gemini umbrella ship on a parallel cadence to the chat models. Imagen 1 (May 2022) was a Google Brain research model; Imagen 2 (December 2023) powered Bard's image generation and was the model behind the February 2024 incident; Imagen 3 (announced May 14, 2024 at I/O; ImageFX rollout August 15, 2024; people-image return August 28, 2024) introduced the photorealistic-rendering jump. Imagen 4 shipped at I/O 2025 (May 20, 2025) in Standard, Fast, and Ultra variants with major text-rendering improvements and up to 2K resolution; Workspace Docs / Slides / Vids integration followed.

Veo 1 debuted as a Vertex research preview at I/O 2024. Veo 2 (December 16, 2024) brought 4K resolution and clips up to two minutes. Veo 3 at I/O 2025 was the first Veo with native synchronized audio — dialogue, sound effects, and ambient sound generated in the same pass — available initially through the AI Ultra tier and via the Google Flow filmmaking surface. Image and video models are deliberately not enumerated as separate rows in the version table above; if either line grows substantially, a dedicated /ai/gemini/image-models/ sibling page is in scope.

People who shaped Gemini

Sundar Pichai — CEO of Google since 2015, CEO of Alphabet since December 2019. The 2023 ChatGPT-response framing, the Brain × DeepMind merger, the December 2023 Gemini 1.0 launch, the February 2024 image-generation memo, and the strategy around the DOJ search remedies all run through Pichai's office.

Demis Hassabis — CEO of Google DeepMind since the April 2023 merger. DeepMind cofounder. 2024 Nobel laureate in Chemistry (jointly with John Jumper of DeepMind and David Baker) for AlphaFold. The technical-leadership face of the post-merger Gemini program.

Jeff Dean — Chief Scientist, Google + Alphabet (since the April 2023 merger), reporting directly to Pichai. Previously co-led Google Brain. Sissie Hsiao led the Bard product through the rename. Eli Collins (VP, Google DeepMind product) co-launched Bard alongside Hsiao. Prabhakar Raghavan (then SVP, Search) authored the February 2024 image-generation explainer; he transitioned out of Search leadership in October 2024.

Current leadership beyond Pichai / Hassabis / Dean shifts more frequently and should be re-verified at each refresh run; the Musk v. Altman–style governance episode that has played out at OpenAI does not have a Google analog.

The competitive landscape

Gemini competes most directly with OpenAI's ChatGPT (the line that Bard was Google's reactive response to in 2023; see ChatGPT Versions), Anthropic's Claude (Google's primary capital and cloud partner alongside Amazon; see Claude Versions), xAI's Grok (see Grok Versions), and Meta's Llama. Gemini's distinguishing positioning has been the integration into Search, Workspace, Pixel devices, and Chrome alongside the standalone Gemini app, and the on-device Nano line that no other major frontier lab has matched at scale. This page does not attempt a benchmark roundup or a ranking.

Use Gemini

The browser cannot detect which Gemini model you've used or are using — there's no fingerprint or header that exposes it. The block below carries the practical information instead: the current model strings, a copy-paste API call, and the surfaces where Gemini is available.

Current model strings

Use these in the model field of an API request. Verify against ai.google.dev/gemini-api/docs/models for the freshest list.

# Flagship Pro (preview) — current Google recommendation
gemini-3.1-pro-preview

# Flash — default in the Gemini app and Search AI Mode
gemini-3-flash-preview

# 2.5 family — stable GA, still routinely served
gemini-2.5-pro
gemini-2.5-flash
gemini-2.5-flash-lite

# Specialized — image generation and on-device
gemini-2.5-flash-image
gemini-nano   # on-device via AICore / ML Kit GenAI APIs

Quick API call

Drop in your GEMINI_API_KEY and run. The generateContent endpoint is the canonical entry point.

$ curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-3.1-pro-preview:generateContent" \
    -H "x-goog-api-key: $GEMINI_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "contents": [{
        "parts": [{ "text": "Hello, Gemini." }]
      }]
    }'

Where to access Gemini

Multiple surfaces, same models underneath. Pick whichever fits the task.

# Standalone consumer app — free + Google AI Pro / Ultra
https://gemini.google.com/

# Developer surfaces
https://ai.google.dev/                         # Gemini API + AI Studio
https://aistudio.google.com/
https://cloud.google.com/vertex-ai/             # Vertex AI (enterprise)

# Built-in surfaces
Search (AI Mode + AI Overviews)
Workspace (Gmail, Docs, Sheets, Slides, Meet)   # formerly Duet AI
Pixel + Android (on-device Gemini Nano)
Chrome (Built-in AI Prompt API powered by Nano)

# Native apps
Gemini for iOS, Android

Model lifecycle

Google publishes a changelog alongside each new model id; preview ids are eventually replaced with stable equivalents. Pin a dated id only when you need bit-for-bit reproducibility; otherwise prefer the un-dated alias so you migrate forward automatically.

# Changelog and lifecycle
https://ai.google.dev/gemini-api/docs/changelog

# Stable alias — rolls forward to the latest snapshot
"model": "gemini-2.5-pro"

# Pinned snapshot — freezes the exact training cut
"model": "gemini-2.5-pro-preview-06-05"