AI Radar

Google models

Google, headquartered in Mountain View, California, makes hundreds of products, among them Google Search, YouTube, Android and Gmail. Its AI research unit, Google DeepMind, was formed in 2023 by combining Google Brain and DeepMind. Google's language models include LaMDA, PaLM and the Gemini line, and since 2024 it has also released open-weights models under the Gemma name.

Chronology

  1. 2026

    10 models
    1. Announced, release date unknown
      Gemini · Language, Multimodal, Reasoning · Milestone Gemini 4 Argon

      Gemini 4 Argon is a reasoning model in Google's Gemini line, announced on 30 September 2026 and first rolled out only to vetted cyber defenders in the Fairwind Program. Google describes it as built…

      Primary source Preview · 2 news items
    2. Released
      Gemini Flash · Language, Multimodal, Reasoning Gemini 3.8 Flash

      Gemini 3.8 Flash is a multimodal reasoning model in Google's Flash line with adjustable effort levels, released on 2 September 2026 and based on Gemini 3.7 Flash. Google aims it at long-horizon…

      • Takes extra reasoning steps and calls tools repeatedly on complex tasks, so Google says it can use more tokens at higher effort levels; 3.7 Flash stays supported.
      • Google says software engineering, agentic tasks and multi-step reasoning in specialised domains improved over 3.7 Flash.
      Primary source Available · 2 news items
    3. Released
      Gemini Flash · Language, Multimodal, Reasoning Gemini 3.7 Flash

      Gemini 3.7 Flash is a multimodal reasoning model in Google's Gemini Flash line, released on 13 August 2026 and based on Gemini 3.6 Flash. Google positions it for coding and agents; it was offered…

      • Algorithmic changes to its core reasoning; Google says it puts more effort into multi-step planning and plans tool calls more carefully.
      • Introductory price of 0.75 USD per million input tokens and 3.75 USD per million output tokens through 31 December 2026, half the original 3.6 Flash price.
      Primary source Available · 1 news item
    4. Released
      Gemini Flash-Lite · Language, Multimodal, Reasoning Gemini 3.5 Flash-Lite

      Gemini 3.5 Flash-Lite is a multimodal reasoning model in Google's Flash-Lite line, released on 21 July 2026 through the Gemini API, Gemini Enterprise Agent Platform and the Gemini app. Based on…

      • Computer use is available as a built-in tool; the Gemini API model page lists it as a preview capability.
      • Google reports higher quality than 3.1 Flash-Lite across thinking levels in coding, agentic, long-context and real-world task evaluations.
      Primary source Available · 1 news item
    5. Released
      Gemini Flash · Language, Multimodal, Reasoning Gemini 3.6 Flash

      Gemini 3.6 Flash is a multimodal reasoning model in Google's Gemini Flash line, released on 21 July 2026 through the Gemini API, Google Antigravity, Gemini Enterprise and the Gemini app. Google…

      • Google says it uses fewer output tokens, reasoning steps and tool calls than 3.5 Flash to complete multi-step workflows.
      • Launched at 1.50 USD per million input tokens and 7.50 USD per million output tokens, with a lower output token price than 3.5 Flash.
      Primary source Available · 1 news item
    6. Released
      Gemini Flash · Language, Multimodal, Reasoning Gemini 3.5 Flash

      Gemini 3.5 Flash is a multimodal language model in Google's Gemini Flash line and the first model of the Gemini 3.5 series, generally available from 19 May 2026 in the Gemini API, Agent Platform…

      • Generally available as a stable model from release, whereas Gemini 3 Flash remained a preview model.
      • The default thinking effort is medium instead of high.
      Primary source Available · 2 news items
    7. Announced, release date unknown
      Gemini Pro · Language Gemini 3.5 Pro

      Gemini 3.5 Pro is an announced language model in Google's Gemini Pro line, named on 19 May 2026 as already in internal use, with a rollout planned for the following month. In July 2026 Google said…

      Primary source Announced · no news linked
    8. Released
      Gemma · Language, Multimodal, Reasoning · Milestone Gemma 4

      Gemma 4 is a family of open-weight models from Google, released on 2 April 2026 under the Apache 2.0 licence in E2B, E4B, 26B Mixture-of-Experts and 31B dense sizes, with a 12B Unified size added in…

      • Released under the Apache 2.0 licence, a choice Google links to developer feedback.
      • Adds a configurable thinking mode in all sizes, which produces a reasoning trace before the answer.
      Primary source Available · 1 news item
    9. Released
      Gemini Flash-Lite · Language, Multimodal, Reasoning Gemini 3.1 Flash-Lite

      Gemini 3.1 Flash-Lite is a multimodal language model and the first Flash-Lite model of Google's Gemini 3 series, released in preview on 3 March 2026 in the Gemini API and Vertex AI and generally…

      • Based on Gemini 3 Pro, according to Google's model card.
      Primary source Available · no news linked
    10. Released
      Gemini Pro · Language, Multimodal, Reasoning Gemini 3.1 Pro

      Gemini 3.1 Pro is a multimodal language model in Google's Gemini Pro line, released in preview on 19 February 2026 in the Gemini API, Vertex AI, the Gemini app and NotebookLM. Google calls it the…

      • Adds MEDIUM as a thinking level.
      • Offers a separate endpoint, gemini-3.1-pro-preview-customtools, tuned to prioritise custom tools for developers who mix bash and custom tools.
      Primary source Preview · no news linked

  2. 2025

    9 models
    1. Released
      Gemini Flash · Language, Multimodal, Reasoning Gemini 3 Flash

      Gemini 3 Flash is a multimodal language model in Google's Gemini Flash line, released in preview on 17 December 2025 through the Gemini API and Vertex AI and as the default model in the Gemini app…

      • Replaced Gemini 2.5 Flash as the default model in the Gemini app and began rolling out as the default model for AI Mode in Search.
      • Based on Gemini 3 Pro, according to Google's model card, which says it builds on that model's reasoning foundation.
      Primary source Preview · no news linked
    2. Released
      Gemini Pro · Language, Multimodal, Reasoning · Milestone Gemini 3 Pro

      Gemini 3 Pro is a multimodal reasoning model in Google's Gemini Pro line and the first Gemini 3 model, released in preview on 18 November 2025 in the Gemini app, AI Mode in Search, Google AI Studio…

      • Adds an optional Deep Think mode, meant to improve complex problem solving at inference time.
      • Available in AI Mode in Google Search on launch day, which Google says was a first for a Gemini model.
      Primary source Retired · no news linked
    3. Released
      Gemma · Language, Multimodal Gemma 3n

      Gemma 3n is a family of open-weight models from Google for on-device use, previewed on 20 May 2025 and fully released on 26 June 2025 in E2B and E4B sizes. Google presents it as a mobile-first…

      • New mobile-first architecture with MatFormer nesting, Per-Layer Embeddings and KV cache sharing.
      • Adds audio input, for speech recognition and speech translation, next to text, image and video.
      Primary source Available · no news linked
    4. Released
      Gemini Flash-Lite · Language, Multimodal, Reasoning Gemini 2.5 Flash-Lite

      Gemini 2.5 Flash-Lite is a multimodal reasoning model in Google's Gemini Flash-Lite line, released in preview on 17 June 2025 and generally available in the Gemini API from 22 July 2025. It has a 1…

      • Adds thinking with a configurable budget, off by default.
      • Can connect to tools such as Google Search and code execution.
      Primary source Available · no news linked
    5. Released
      Gemini Flash · Language, Multimodal, Reasoning Gemini 2.5 Flash

      Gemini 2.5 Flash is a multimodal reasoning model in Google's Gemini Flash line, released in preview on 17 April 2025 and generally available from 17 June 2025 in the Gemini API and the Gemini app…

      • Adds controllable thinking: it can be switched on or off and capped with a thinking budget.
      • Google says it improves reasoning over 2.0 Flash while keeping that model's speed when thinking is off.
      Primary source Available · no news linked
    6. Released
      Gemini Pro · Language, Multimodal, Reasoning Gemini 2.5 Pro

      Gemini 2.5 Pro is a multimodal reasoning model in Google's Gemini Pro line, first released as an experimental model on 25 March 2025 and generally available from 17 June 2025 in the Gemini API and…

      • Built as a thinking model that reasons through its thoughts before responding; Google says it will build thinking into all its models.
      • Google reports a large step up in coding performance over Gemini 2.0.
      Primary source Available · no news linked
    7. Released
      Gemma · Language, Multimodal · Milestone Gemma 3

      Gemma 3 is a family of open-weight models from Google, released on 12 March 2025 in 1B, 4B, 12B and 27B sizes, with a 270M size added in August 2025. Google says it is built from the same research…

      • Adds image input for the 4B, 12B and 27B sizes through a shared vision encoder; the 1B size stays text-only.
      • Context window raised to 128K tokens for the 4B, 12B and 27B sizes and to 32K for the 1B size.
      Primary source Available · no news linked
    8. Released
      Gemini Flash-Lite · Language, Multimodal Gemini 2.0 Flash-Lite

      Gemini 2.0 Flash-Lite is a multimodal language model from Google, released in public preview on 5 February 2025 and generally available in the Gemini API, through Google AI Studio and Vertex AI…

      • Google says it offers better quality than Gemini 1.5 Flash at the same speed and cost.
      • Uses a single price per input type, dropping the split between short and long context prices used for Gemini 1.5 Flash.
      Primary source Retired · no news linked
    9. Released
      Gemini Pro · Language, Multimodal Gemini 2.0 Pro

      Gemini 2.0 Pro is a multimodal language model in Google's Gemini Pro line, released on 5 February 2025 only as an experimental model, in Google AI Studio, Vertex AI and the Gemini app for Gemini…

      • Google says it improved coding and the handling of complex prompts, world knowledge and reasoning compared with the models it had released before.
      Primary source Retired · no news linked

  3. 2024

    11 models
    1. Released
      Gemini Flash · Language, Reasoning Gemini 2.0 Flash Thinking

      Gemini 2.0 Flash Thinking is an experimental reasoning model from Google, built on Gemini 2.0 Flash and offered in preview in the Gemini API and Google AI Studio from December 2024 and in the Gemini…

      • Adds a thinking process: Google says the model is trained to break prompts into a series of steps, and it shows its reasoning before answering.
      Primary source Retired · no news linked
    2. Released
      Gemini Flash · Language, Multimodal · Milestone Gemini 2.0 Flash

      Gemini 2.0 Flash is the first model of Google's Gemini 2.0 family, released as an experimental model in December 2024 and generally available from February 2025 through the Gemini API, Google AI…

      • Adds native image and audio output, first limited to early-access partners; at general availability output was text only.
      • Natively calls tools such as Google Search and code execution, alongside user-defined functions.
      Primary source Retired · no news linked
    3. Released
      Gemini Flash · Language, Multimodal Gemini-1.5-Flash-002

      Gemini-1.5-Flash-002 is an updated, production-ready version of Google's Gemini 1.5 Flash, released with Gemini-1.5-Pro-002 through Google AI Studio, the Gemini API and Vertex AI. Google says it…

      • Gives more concise answers by default, which Google says makes the model easier and cheaper to use.
      • Safety filters are no longer applied by default; developers choose the configuration.
      Primary source Retired · no news linked
    4. Released
      Gemini Pro · Language, Multimodal Gemini-1.5-Pro-002

      Gemini-1.5-Pro-002 is an updated, production-ready version of Google's Gemini 1.5 Pro, released with Gemini-1.5-Flash-002 through Google AI Studio, the Gemini API and Vertex AI, with a 2 million…

      • Gives more concise answers by default, which Google says makes the model easier and cheaper to use.
      • Safety filters are no longer applied by default; developers choose the configuration.
      Primary source Retired · no news linked
    5. Released
      Gemini Flash · Language, Multimodal Gemini 1.5 Flash-8B

      Gemini 1.5 Flash-8B is a smaller and faster variant of Google's Gemini 1.5 Flash, first released as an experimental API model in August 2024 and production-ready in Google AI Studio and the Gemini…

      • A smaller model than 1.5 Flash; the technical report says it inherits the core architecture, optimizations and data mixture of 1.5 Flash.
      • Lower latency on small prompts, and at its stable release half the price and twice the rate limits of 1.5 Flash.
      Primary source Retired · no news linked
    6. Released
      Gemma · Language · Milestone Gemma 2

      Gemma 2 is the second generation of Google's Gemma open-weight language models, released in 9B and 27B sizes with weights on Kaggle and Hugging Face, followed by a 2B size in July 2024. Google says…

      • Alternates local sliding-window and global attention layers and adds grouped-query attention and logit soft-capping, with deeper networks.
      • The 2B and 9B models are trained with knowledge distillation from a larger teacher model instead of next-token prediction; the 27B model is trained from scratch.
      Primary source Available · no news linked
    7. Released
      Gemini Pro · Language, Multimodal Gemini 1.5 Pro

      Gemini 1.5 Pro is a mid-size multimodal model in Google's Gemini Pro line and the first model of the Gemini 1.5 generation, announced in February 2024 and built on a Mixture-of-Experts architecture…

      • Uses a Mixture-of-Experts architecture, which Google says makes the model more efficient to train and serve.
      • Standard context window grows from 32,000 to 128,000 tokens, with up to 1 million in private preview and 2 million generally available in the Gemini API from June 2024.
      Primary source Retired · no news linked
    8. Released
      Gemini Flash · Language, Multimodal Gemini 1.5 Flash

      Gemini 1.5 Flash is a Google model that accepts text, images, audio and video and returns text, introduced at Google I/O 2024 in public preview in Google AI Studio and Vertex AI with a 1 million…

      • A smaller, lighter-weight model than 1.5 Pro.
      • Designed for lower latency and lower cost to serve than 1.5 Pro, according to Google.
      Primary source Retired · no news linked
    9. Released
      Gemma · Language Gemma 1.1

      Gemma 1.1 is an update to Google's first-generation Gemma open models, released as instruction-tuned 2B and 7B checkpoints with downloadable weights under Google's Gemma terms of use. Google…

      • Post-trained with a new RLHF method; Google says this improves quality, coding, factuality, instruction following and multi-turn conversation.
      • Fixes a bug in multi-turn conversations, and responses no longer always open with the same word, according to the model card.
      Primary source Available · no news linked
    10. Released
      Gemma · Language · Milestone Gemma

      Gemma is a family of lightweight open-weight language models from Google, released in February 2024 in 2B and 7B sizes, each pre-trained and instruction-tuned, and built from the research behind…

      Primary source Available · no news linked
    11. Released
      Gemini Ultra · Language, Multimodal · Milestone Gemini 1.0 Ultra

      Gemini 1.0 Ultra is the biggest of the three sizes of Google's Gemini 1.0 models, which Google describes as natively multimodal, announced in December 2023. It reached consumers as Ultra 1.0 in…

      Primary source Status unknown · no news linked Some details lack sources

  4. 2023

    3 models
    1. Released
      Gemini Nano · Language, Multimodal · Milestone Gemini 1.0 Nano

      Gemini 1.0 Nano is the on-device size of Google's Gemini 1.0 models, distilled from larger Gemini models and 4-bit quantized, in versions of 1.8B and 3.25B parameters. It first ran on the Pixel 8…

      Primary source Status unknown · no news linked Some details lack sources
    2. Released
      Gemini Pro · Language, Multimodal · Milestone Gemini 1.0 Pro

      Gemini 1.0 Pro is the Gemini 1.0 size Google meant for a wide range of tasks. A tuned version ran in Bard from December 2023, and a week later it reached developers through the Gemini API and Vertex…

      Primary source Retired · no news linked
    3. Released
      PaLM · Language · Milestone PaLM 2

      PaLM 2 is Google's second-generation PaLM language model, announced at Google I/O in May 2023 in four sizes named Gecko, Otter, Bison and Unicorn, and offered the same day in Bard, through the PaLM…

      • Pre-trained much more heavily on multilingual text spanning more than 100 languages; the data mix also adds more code, mathematics and parallel multilingual documents.
      • Trained with a tuned mixture of pre-training objectives, following UL2; Google says this makes it more compute-efficient than PaLM, with faster inference.
      Primary source Retired · no news linked

  5. 2022

    2 models
    1. Released
      LaMDA · Language · Milestone LaMDA 2

      LaMDA 2 is the second version of Google's conversational language model, announced at Google I/O in May 2022. Google offered it only through AI Test Kitchen, an app of experimental demos that began…

      • Google says testing with thousands of its employees brought quality gains and fewer inaccurate or offensive responses, and that the demo version had key safety improvements.
      • Opened to a wider audience through experimental demos in the AI Test Kitchen app, after earlier testing inside Google.
      Primary source Status unknown · no news linked Some details lack sources
    2. Announced, release date unknown
      PaLM · Language · Milestone PaLM

      PaLM (Pathways Language Model) is a Google research language model, introduced in April 2022 as a 540-billion-parameter dense decoder-only Transformer trained with Google's Pathways system on 6,144…

      Primary source Research only · no news linked

  6. 2021

    1 model
    1. Announced, release date unknown
      LaMDA · Language · Milestone LaMDA

      LaMDA (Language Model for Dialogue Applications) is a Google research language model trained on dialogue, announced in May 2021; its main version has 137 billion parameters. The public could use it…

      Primary source Retired · no news linked

Lineage

  1. LaMDA
    Successors
    LaMDA 2
  2. PaLM
    Successors
    PaLM 2
  3. LaMDA 2
    Predecessors
    LaMDA
  4. PaLM 2
    Predecessors
    PaLM
  5. Gemini 1.0 Nano

    No recorded relations.

  6. Gemini 1.0 Pro
    Successors
    Gemini 1.5 Pro
  7. Gemini 1.0 Ultra

    No recorded relations.

  8. Gemma
    Successors
    Gemma 1.1
    Variants and derived
    Zephyr 7B Gemma (Hugging Face) · derived: fine tune
  9. Gemma 1.1
    Predecessors
    Gemma
    Successors
    Gemma 2
  10. Gemini 1.5 Flash
    Based on
    Gemini 1.5 Pro · derived: distillation
    Variants and derived
    Gemini-1.5-Flash-002 · revision; Gemini 1.5 Flash-8B · variant
  11. Gemini 1.5 Pro
    Predecessors
    Gemini 1.0 Pro
    Variants and derived
    Gemini 1.5 Flash · derived: distillation; Gemini-1.5-Pro-002 · revision
  12. Gemma 2
    Predecessors
    Gemma 1.1
    Successors
    Gemma 3
  13. Gemini 1.5 Flash-8B
    Based on
    Gemini 1.5 Flash · variant
  14. Gemini-1.5-Flash-002
    Based on
    Gemini 1.5 Flash · revision
  15. Gemini-1.5-Pro-002
    Successors
    Gemini 2.0 Pro
    Based on
    Gemini 1.5 Pro · revision
  16. Gemini 2.0 Flash
    Variants and derived
    Gemini 2.0 Flash Thinking · derived: other
  17. Gemini 2.0 Flash Thinking
    Based on
    Gemini 2.0 Flash · derived: other
  18. Gemini 2.0 Flash-Lite
  19. Gemini 2.0 Pro
    Predecessors
    Gemini-1.5-Pro-002
    Successors
    Gemini 2.5 Pro
  20. Gemma 3
    Predecessors
    Gemma 2
    Successors
    Gemma 4
  21. Gemini 2.5 Pro
    Predecessors
    Gemini 2.0 Pro
    Successors
    Gemini 3 Pro
  22. Gemini 2.5 Flash
    Predecessors
    Gemini 2.0 Flash
    Successors
    Gemini 3 Flash
  23. Gemini 2.5 Flash-Lite
  24. Gemma 3n

    No recorded relations.

  25. Gemini 3 Pro
    Predecessors
    Gemini 2.5 Pro
    Successors
    Gemini 3.1 Pro
    Variants and derived
    Gemini 3.1 Flash-Lite · derived: other; Gemini 3 Flash · derived: other
  26. Gemini 3 Flash
    Predecessors
    Gemini 2.5 Flash
    Based on
    Gemini 3 Pro · derived: other
  27. Gemini 3.1 Pro
    Predecessors
    Gemini 3 Pro
  28. Gemini 3.1 Flash-Lite
    Based on
    Gemini 3 Pro · derived: other
  29. Gemma 4
    Predecessors
    Gemma 3
  30. Gemini 3.5 Flash
    Predecessors
    Gemini 3 Flash
  31. Gemini 3.5 Pro

    No recorded relations.

  32. Gemini 3.5 Flash-Lite
  33. Gemini 3.6 Flash
    Predecessors
    Gemini 3.5 Flash
  34. Gemini 3.7 Flash
    Predecessors
    Gemini 3.6 Flash
  35. Gemini 3.8 Flash
    Predecessors
    Gemini 3.7 Flash
  36. Gemini 4 Argon

    No recorded relations.

Families

About the organisation

  • Description Google, headquartered in Mountain View, California, makes hundreds of products, among them Google Search, YouTube, Android and Gmail. Its AI research unit, Google DeepMind, was formed in 2023 by combining Google Brain and DeepMind. Google's language models include LaMDA, PaLM and the Gemini line, and since 2024 it has also released open-weights models under the Gemma name. Primary source[1] [2] [3] [4] [5] [6] [7]

Company · https://about.google/ · 36 models listed here

Sources

  1. [1] From the garage to the Googleplex

    Other · Google

    Provenance
    Primary
    Availability
    Active
    Last checked
    Original ↗ https://about.google/company-info/our-story/ About Google company page. Title is the page heading; the HTML title reads How we started and where we are today, and the section is labelled Our story. No publication date stated. The old address https://about.google/our-story/ redirects here.

  2. [2] Google DeepMind: Bringing together two world-class AI teams

    Announcement · Google · published

    Provenance
    Primary
    Availability
    Active
    Last checked
    Original ↗ https://blog.google/innovation-and-ai/technology/ai/april-ai-update/ Originally published at https://blog.google/technology/ai/april-ai-update/, which now redirects here.

  3. [3] About Google DeepMind

    Other · Google

    Provenance
    Primary
    Availability
    Active
    Last checked
    Original ↗ https://deepmind.google/about/ Living page; content as seen on 2026-10-01. Title from the HTML title; the page heading is the mission statement. No publication date stated.

  4. [4] LaMDA: our breakthrough conversation technology

    Announcement · Google · published

    Provenance
    Primary
    Availability
    Active
    Last checked
    Original ↗ https://blog.google/innovation-and-ai/products/lamda/ The old address https://blog.google/technology/ai/lamda/ redirects (301) here.

  5. [5] Pathways Language Model (PaLM): Scaling to 540 Billion Parameters for Breakthrough Performance

    Blog post · Google · published

    Provenance
    Primary
    Availability
    Active
    Last checked
    Original ↗ https://research.google/blog/pathways-language-model-palm-scaling-to-540-billion-parameters-for-breakthrough-performance/ Google Research blog post announcing the PaLM paper (arXiv 2204.02311).

  6. [6] Introducing Gemini: our largest and most capable AI model

    Announcement · Google · published

    Provenance
    Primary
    Availability
    Active
    Last checked
    Original ↗ https://blog.google/innovation-and-ai/technology/ai/google-gemini-ai/ Headline as shown on the page; the HTML title differs (Introducing Gemini: Google's most capable AI model yet). Originally published at https://blog.google/technology/ai/google-gemini-ai/, which now redirects here.

  7. [7] Gemma: Introducing new state-of-the-art open models

    Announcement · Google · published

    Provenance
    Primary
    Availability
    Active
    Last checked
    Original ↗ https://blog.google/innovation-and-ai/technology/developers-tools/gemma-open-models/ Headline as shown on the page; the HTML title is Gemma: Google introduces new state-of-the-art open models. The old address https://blog.google/technology/developers/gemma-open-models/ redirects (301) here.