---
title: "Google’s large language models"
source: "https://systems-analysis.info/eng/Google%E2%80%99s_large_language_models"
wiki: "systems-analysis.info/eng"
article: "Google’s_large_language_models"
language: "en"
categories:
  - "Category:English"
  - "Category:Google"
  - "Category:Large language models"
  - "Category:LLM families"
  - "Category:Machine learning"
  - "Category:Technology"
revision_id: 169
wiki_created_at: 2026-09-06T22:18:22Z
wiki_modified_at: 2026-09-06T22:18:22Z
downloaded_at: 2026-09-07T22:21:32Z
---

# Google’s large language models

**Google's [large language models](https://systems-analysis.info/eng/Large_language_model "Large language model")** are a series of large language models (LLMs) developed by various divisions within Google, including Google AI (formerly Google Brain) and DeepMind. As one of the pioneers in deep learning and the [Transformer architecture](https://systems-analysis.info/eng/Transformer_architecture "Transformer architecture"), Google has made fundamental contributions to the development of modern LLMs. The history of these models' development reflects a journey from specialized language understanding systems to large-scale [multimodal](https://systems-analysis.info/eng/Multimodal_large_language_models "Multimodal large language models") and [agentic](https://systems-analysis.info/eng/Agentic_Workflows "Agentic Workflows") systems that underpin many of Google's products and define the trajectory of the entire AI industry.

## History and evolution of Google's models

### Early achievements and neural machine translation (2011–2016)

The foundations for LLM development at Google were laid by the **Google Brain** project (2011), which was dedicated to the application of deep neural networks. One of the first breakthroughs was the **Word2Vec** algorithm (2013), created by Tomas Mikolov. It allowed words to be represented as vectors ([embeddings](https://systems-analysis.info/eng/Embedding_(NLP) "Embedding (NLP)")) that captured their semantic context, which became a fundamental method for language understanding in neural networks.

The next step was the shift to sequence-to-sequence models, such as **seq2seq** (2014), which formed the basis of **Google Neural Machine Translation (GNMT)** (2016). The transition of Google Translate to an LSTM-based neural network architecture significantly improved the quality of machine translation. In parallel, the subsidiary DeepMind, acquired by Google in 2014, demonstrated the power of deep learning with the victory of the **AlphaGo** system over the world Go champion, solidifying belief in the potential of AI.

### The Transformer revolution and the birth of BERT (2017–2018)

In 2017, researchers at Google Brain introduced the **Transformer** architecture in the paper "Attention Is All You Need." This architecture, based on the **self-attention** mechanism, allowed sequences to be processed in parallel rather than sequentially, which became a revolution in NLP and the foundation for all modern LLMs.

Riding this wave of success, Google introduced the **[BERT](https://systems-analysis.info/eng/BERT_(language_model) "BERT (language model)")** (Bidirectional [Encoder](https://systems-analysis.info/eng/Encoder_(Transformer) "Encoder (Transformer)") Representations from Transformers) model in 2018. BERT was the first deeply **bidirectional** model, which considered the context of a word from both the left and the right simultaneously. This enabled it to achieve state-of-the-art results on numerous language understanding tasks ([GLUE](https://systems-analysis.info/eng/GLUE_Benchmark "GLUE Benchmark"), SQuAD) and set a new industry standard. BERT was released in two versions (BASE with 110 million parameters and LARGE with 340 million) with open source code and weights, which contributed to its widespread adoption. Since 2019, BERT has been used in Google Search to better understand user queries.

### Increasing scale and the era of conversational models (2019–2022)

After BERT, Google continued to experiment with scale and architecture:

- **[T5](https://systems-analysis.info/eng/T5_(Text-to-Text_Transfer_Transformer) "T5 (Text-to-Text Transfer Transformer)") (Text-to-Text Transfer Transformer, 2019)**: A unified model that treats every NLP task as a "text-to-text" problem. Trained on the massive C4 (Colossal Clean Crawled Corpus), T5 was also released as open source in several sizes (up to 11 billion parameters).
- **Meena (2020)**: Google's first specialized conversational model with 2.6 billion parameters, which demonstrated high-quality open-domain dialogue.
- **[LaMDA](https://systems-analysis.info/eng/LaMDA_(Google) "LaMDA (Google)") (Language Model for Dialogue Applications, 2021)**: A family of conversational models (up to 137 billion parameters) trained on a vast corpus of dialogues (1.56 trillion words). LaMDA was aimed at creating more natural and meaningful conversations and became widely known after a Google engineer claimed it was "sentient."
- **Gopher and [Chinchilla](https://systems-analysis.info/eng/Chinchilla_(language_model) "Chinchilla (language model)") (DeepMind, 2021–2022)**: In parallel, DeepMind investigated scaling laws. The **Gopher** model (280 billion parameters) showed how scale affects quality. And the **Chinchilla** model (70 billion) demonstrated that for optimal performance, the key is not the maximum number of parameters but the correct balance between model size and the amount of training data. This finding became known as "Chinchilla's law" and influenced LLM training strategies across the industry.

### The era of massive and multimodal models (2022–present)

- **[PaLM](https://systems-analysis.info/eng/PaLM_(Pathways_Language_Model) "PaLM (Pathways Language Model)") (Pathways Language Model, 2022)**: At the time of its announcement, it was Google's largest dense model, with **540 billion parameters**, trained on the new Pathways distributed infrastructure. PaLM demonstrated breakthrough capabilities in logical reasoning, especially using the **[Chain-of-Thought](https://systems-analysis.info/eng/Chain-of-Thought_Prompting "Chain-of-Thought Prompting") (CoT) prompting** technique. Specialized versions were built on it, such as **Med-PaLM** for medicine. In 2023, an improved version, **PaLM 2** (~340 billion parameters), was released, which became the foundation for the updated Bard chatbot.
- **[Gemini](https://systems-analysis.info/eng/Gemini_(Google) "Gemini (Google)") (2023–present)**: A new generation of models created by the combined Google DeepMind team. Gemini was designed from the ground up as a **natively multimodal** system, capable of processing text, code, images, audio, and video. It was released in several versions:
  - **Gemini Ultra**: The most powerful model for complex tasks.
  - **Gemini Pro**: A versatile model for a wide range of tasks.
  - **Gemini Nano**: A compact model for on-device operation on mobile devices.

In 2024–2025, the family was expanded with **Gemini 1.5** (with a [context window](https://systems-analysis.info/eng/Context_window "Context window") of up to 1 million [tokens](https://systems-analysis.info/eng/Token_(LLM) "Token (LLM)")) and **Gemini 2.0**, which gained agentic capabilities.

## Summary table of key models

| Model                                                                                                                       | Release year | Release date | Parameters (est.)                 | Architecture                                 | Key features                                                                                                                | arXiv paper                                                                                          | Stable Google source                                                                                                                                                                              |
|-----------------------------------------------------------------------------------------------------------------------------|--------------|--------------|-----------------------------------|----------------------------------------------|-----------------------------------------------------------------------------------------------------------------------------|------------------------------------------------------------------------------------------------------|---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| **[Transformer](https://systems-analysis.info/eng/Transformer_architecture "Transformer architecture")**                    | 2017         | 2017-06-12   | ~65–213 million                   | Encoder-Decoder                              | Self-attention sequence model; foundational architecture for modern language models.                                        | <a href="https://arxiv.org/abs/1706.03762" class="external text" rel="nofollow">arXiv:1706.03762</a> | <a href="https://research.google/pubs/attention-is-all-you-need/" class="external text" rel="nofollow">Google Research</a>                                                                        |
| **[BERT](https://systems-analysis.info/eng/BERT_(language_model) "BERT (language model)")**                                 | 2018         | 2018-11-02   | 110–340 million                   | Encoder                                      | Bidirectional masked-language pre-training; major breakthrough in language understanding tasks.                             | <a href="https://arxiv.org/abs/1810.04805" class="external text" rel="nofollow">arXiv:1810.04805</a> | <a href="https://research.google/blog/open-sourcing-bert-state-of-the-art-pre-training-for-natural-language-processing/" class="external text" rel="nofollow">Google Research Blog</a>            |
| **XLNet**                                                                                                                   | 2019         | 2019-06-19   | 110–340 million                   | Autoregressive Transformer-XL                | Permutation language modeling; bidirectional context without BERT-style masking mismatch.                                   | <a href="https://arxiv.org/abs/1906.08237" class="external text" rel="nofollow">arXiv:1906.08237</a> | <a href="https://research.google/blog/google-at-neurips-2019/" class="external text" rel="nofollow">Google Research Blog</a>                                                                      |
| **ALBERT**                                                                                                                  | 2019         | 2019-12-20   | ~12–235 million                   | Encoder                                      | Parameter-sharing and factorized embeddings; more efficient BERT-style pre-training.                                        | <a href="https://arxiv.org/abs/1909.11942" class="external text" rel="nofollow">arXiv:1909.11942</a> | <a href="https://research.google/blog/albert-a-lite-bert-for-self-supervised-learning-of-language-representations/" class="external text" rel="nofollow">Google Research Blog</a>                 |
| **[T5](https://systems-analysis.info/eng/T5_(Text-to-Text_Transfer_Transformer) "T5 (Text-to-Text Transfer Transformer)")** | 2019         | 2019-10-23   | 60 million – 11 billion           | Encoder-Decoder                              | Unified "text-to-text" framework for translation, summarization, classification, QA, and other NLP tasks.                   | <a href="https://arxiv.org/abs/1910.10683" class="external text" rel="nofollow">arXiv:1910.10683</a> | <a href="https://research.google/blog/exploring-transfer-learning-with-t5-the-text-to-text-transfer-transformer/" class="external text" rel="nofollow">Google Research Blog</a>                   |
| **Meena**                                                                                                                   | 2020         | 2020-01-28   | 2.6 billion                       | Seq2Seq / Evolved Transformer                | Open-domain multi-turn chatbot; introduced Sensibleness and Specificity Average (SSA) evaluation.                           | <a href="https://arxiv.org/abs/2001.09977" class="external text" rel="nofollow">arXiv:2001.09977</a> | <a href="https://research.google/blog/towards-a-conversational-agent-that-can-chat-aboutanything/" class="external text" rel="nofollow">Google Research Blog</a>                                  |
| **ELECTRA**                                                                                                                 | 2020         | 2020-03-10   | ~14–335 million                   | Encoder + Generator/Discriminator            | Replaced-token detection objective; more sample-efficient than masked-language modeling.                                    | <a href="https://arxiv.org/abs/2003.10555" class="external text" rel="nofollow">arXiv:2003.10555</a> | <a href="https://research.google/blog/more-efficient-nlp-model-pre-training-with-electra/" class="external text" rel="nofollow">Google Research Blog</a>                                          |
| **mT5**                                                                                                                     | 2020         | 2020-10-22   | 300 million – 13 billion          | Encoder-Decoder                              | Multilingual T5 trained across 101 languages.                                                                               | <a href="https://arxiv.org/abs/2010.11934" class="external text" rel="nofollow">arXiv:2010.11934</a> | <a href="https://research.google/pubs/mt5-a-massively-multilingual-pre-trained-text-to-text-transformer/" class="external text" rel="nofollow">Google Research</a>                                |
| **Switch Transformer**                                                                                                      | 2021         | 2021-01-11   | up to 1.6 trillion                | Sparse MoE Encoder-Decoder                   | Simplified expert routing; trillion-parameter sparse scaling with controlled compute cost.                                  | <a href="https://arxiv.org/abs/2101.03961" class="external text" rel="nofollow">arXiv:2101.03961</a> | —                                                                                                                                                                                                 |
| **[LaMDA](https://systems-analysis.info/eng/LaMDA_(Google) "LaMDA (Google)")**                                              | 2021         | 2021-05-18   | up to 137 billion                 | Decoder                                      | Dialogue-specialized Transformer family; focused on open-ended conversation, safety, and factual grounding.                 | <a href="https://arxiv.org/abs/2201.08239" class="external text" rel="nofollow">arXiv:2201.08239</a> | <a href="https://blog.google/innovation-and-ai/products/lamda/" class="external text" rel="nofollow">Google Blog</a>                                                                              |
| **[Gopher](https://systems-analysis.info/eng/Gopher_(Google) "Gopher (Google)")**                                           | 2021         | 2021-12-08   | 280 billion                       | Decoder                                      | DeepMind large-scale language model; broad benchmark, ethics, and retrieval-related analysis.                               | <a href="https://arxiv.org/abs/2112.11446" class="external text" rel="nofollow">arXiv:2112.11446</a> | <a href="https://deepmind.google/blog/language-modelling-at-scale-gopher-ethical-considerations-and-retrieval/" class="external text" rel="nofollow">Google DeepMind Blog</a>                     |
| **RETRO**                                                                                                                   | 2021         | 2021-12-08   | up to 7.5 billion                 | Retrieval-Enhanced Decoder                   | Uses an external retrieval database; competitive performance with far fewer parameters.                                     | <a href="https://arxiv.org/abs/2112.04426" class="external text" rel="nofollow">arXiv:2112.04426</a> | <a href="https://deepmind.google/blog/improving-language-models-by-retrieving-from-trillions-of-tokens/" class="external text" rel="nofollow">Google DeepMind Blog</a>                            |
| **GLaM**                                                                                                                    | 2021         | 2021-12-09   | 1.2 trillion total, sparse active | Sparse MoE Decoder                           | Generalist sparse language model; lower training energy and inference compute than comparable dense models.                 | <a href="https://arxiv.org/abs/2112.06905" class="external text" rel="nofollow">arXiv:2112.06905</a> | <a href="https://research.google/blog/more-efficient-in-context-learning-with-glam/" class="external text" rel="nofollow">Google Research Blog</a>                                                |
| **[Chinchilla](https://systems-analysis.info/eng/Chinchilla_(language_model) "Chinchilla (language model)")**               | 2022         | 2022-03-29   | 70 billion                        | Decoder                                      | Compute-optimal scaling result; showed the importance of training smaller models on more tokens.                            | <a href="https://arxiv.org/abs/2203.15556" class="external text" rel="nofollow">arXiv:2203.15556</a> | <a href="https://deepmind.google/blog/an-empirical-analysis-of-compute-optimal-large-language-model-training/" class="external text" rel="nofollow">Google DeepMind Blog</a>                      |
| **[PaLM](https://systems-analysis.info/eng/PaLM_(Pathways_Language_Model) "PaLM (Pathways Language Model)")**               | 2022         | 2022-04-04   | 540 billion                       | Dense Decoder                                | Pathways-trained language model; strong few-shot reasoning, multilingual, and code capabilities.                            | <a href="https://arxiv.org/abs/2204.02311" class="external text" rel="nofollow">arXiv:2204.02311</a> | <a href="https://research.google/blog/pathways-language-model-palm-scaling-to-540-billion-parameters-for-breakthrough-performance/" class="external text" rel="nofollow">Google Research Blog</a> |
| **UL2**                                                                                                                     | 2022         | 2022-05-10   | up to 20 billion                  | Encoder-Decoder                              | Unified denoising framework combining multiple language-learning objectives.                                                | <a href="https://arxiv.org/abs/2205.05131" class="external text" rel="nofollow">arXiv:2205.05131</a> | <a href="https://research.google/blog/better-language-models-without-massive-compute/" class="external text" rel="nofollow">Google Research Blog</a>                                              |
| **Flan-T5 / Flan-PaLM**                                                                                                     | 2022         | 2022-10-20   | T5: up to 11B; PaLM: 540B         | Instruction-tuned Encoder-Decoder / Decoder  | Instruction tuning and chain-of-thought data; strong zero-shot and few-shot generalization.                                 | <a href="https://arxiv.org/abs/2210.11416" class="external text" rel="nofollow">arXiv:2210.11416</a> | <a href="https://research.google/blog/better-language-models-without-massive-compute/" class="external text" rel="nofollow">Google Research Blog</a>                                              |
| **PaLM 2**                                                                                                                  | 2023         | 2023-05-10   | Not disclosed                     | Transformer                                  | Improved multilingual, reasoning, and coding capabilities; more compute-efficient than PaLM.                                | <a href="https://arxiv.org/abs/2305.10403" class="external text" rel="nofollow">arXiv:2305.10403</a> | <a href="https://blog.google/technology/ai/google-palm-2-ai-large-language-model/" class="external text" rel="nofollow">Google Blog</a>                                                           |
| **Gemini 1.0**                                                                                                              | 2023         | 2023-12-06   | Not disclosed; Nano: 1.8B / 3.25B | Native Multimodal Transformer                | Ultra, Pro, and Nano family; text, image, audio, and video understanding.                                                   | <a href="https://arxiv.org/abs/2312.11805" class="external text" rel="nofollow">arXiv:2312.11805</a> | <a href="https://blog.google/innovation-and-ai/technology/ai/google-gemini-ai/" class="external text" rel="nofollow">Google Blog</a>                                                              |
| **Gemini 1.5**                                                                                                              | 2024         | 2024-02-15   | Not disclosed                     | Multimodal MoE                               | Long-context Gemini generation; up to 1–2 million tokens, with Pro and Flash variants.                                      | <a href="https://arxiv.org/abs/2403.05530" class="external text" rel="nofollow">arXiv:2403.05530</a> | <a href="https://blog.google/innovation-and-ai/products/google-gemini-next-generation-model-february-2024/" class="external text" rel="nofollow">Google Blog</a>                                  |
| **Gemma**                                                                                                                   | 2024         | 2024-02-21   | 2B / 7B                           | Decoder / Open-weight                        | Lightweight open models based on Gemini research; released as pre-trained and instruction-tuned variants.                   | <a href="https://arxiv.org/abs/2403.08295" class="external text" rel="nofollow">arXiv:2403.08295</a> | <a href="https://blog.google/innovation-and-ai/technology/developers-tools/gemma-open-models/" class="external text" rel="nofollow">Google Blog</a>                                               |
| **CodeGemma**                                                                                                               | 2024         | 2024-04-09   | 2B / 7B                           | Decoder / Code-specialized                   | Open code models for code completion, code generation, instruction following, and mathematical reasoning.                   | <a href="https://arxiv.org/abs/2406.11409" class="external text" rel="nofollow">arXiv:2406.11409</a> | <a href="https://deepmind.google/models/gemma/codegemma/" class="external text" rel="nofollow">Google DeepMind</a>                                                                                |
| **PaliGemma**                                                                                                               | 2024         | 2024-05-14   | 3 billion                         | Vision-Language Model                        | Combines SigLIP vision encoder with Gemma language model; captioning, VQA, OCR-like tasks, detection, and segmentation.     | <a href="https://arxiv.org/abs/2407.07726" class="external text" rel="nofollow">arXiv:2407.07726</a> | <a href="https://ai.google.dev/gemma/docs/paligemma" class="external text" rel="nofollow">Google AI for Developers</a>                                                                            |
| **Gemma 2**                                                                                                                 | 2024         | 2024-06-27   | 9B / 27B                          | Decoder / Open-weight                        | Higher-performing and more inference-efficient successor to Gemma; stronger safety tuning.                                  | <a href="https://arxiv.org/abs/2408.00118" class="external text" rel="nofollow">arXiv:2408.00118</a> | <a href="https://blog.google/technology/developers/google-gemma-2/" class="external text" rel="nofollow">Google Blog</a>                                                                          |
| **PaliGemma 2**                                                                                                             | 2024         | 2024-12-04   | 3B / 10B / 28B                    | Vision-Language Model                        | Based on Gemma 2; broader VLM transfer tasks, OCR, object detection, segmentation, and captioning.                          | <a href="https://arxiv.org/abs/2412.03555" class="external text" rel="nofollow">arXiv:2412.03555</a> | <a href="https://deepmind.google/models/gemma/paligemma-2/" class="external text" rel="nofollow">Google DeepMind</a>                                                                              |
| **Gemini 2.0**                                                                                                              | 2024         | 2024-12-11   | Not disclosed                     | Native Multimodal + Tools                    | Agentic-era Gemini line; native image/audio output, native tool use, and stronger multimodal interaction.                   | —                                                                                                    | <a href="https://blog.google/innovation-and-ai/models-and-research/google-deepmind/google-gemini-ai-update-december-2024/" class="external text" rel="nofollow">Google Blog</a>                   |
| **Gemma 3**                                                                                                                 | 2025         | 2025-03-12   | 1B / 4B / 12B / 27B               | Multimodal Open-weight                       | Image-and-text input, text output, 128K context window, and support for 140+ languages.                                     | <a href="https://arxiv.org/abs/2503.19786" class="external text" rel="nofollow">arXiv:2503.19786</a> | <a href="https://blog.google/technology/developers/gemma-3/" class="external text" rel="nofollow">Google Blog</a>                                                                                 |
| **Gemini 2.5**                                                                                                              | 2025         | 2025-03-25   | Not disclosed                     | Multimodal Reasoning Model                   | "Thinking" model family; improved reasoning, coding, mathematics, and complex task performance.                             | —                                                                                                    | <a href="https://deepmind.google/blog/gemini-25-our-world-leading-model-is-getting-even-better/" class="external text" rel="nofollow">Google DeepMind Blog</a>                                    |
| **Gemma 3n**                                                                                                                | 2025         | 2025-06-26   | E2B / E4B                         | On-device Multimodal Open-weight             | Mobile-first Gemma variant; optimized for local multimodal use with efficient memory requirements.                          | —                                                                                                    | <a href="https://developers.googleblog.com/en/introducing-gemma-3n/" class="external text" rel="nofollow">Google Developers Blog</a>                                                              |
| **Gemini 3**                                                                                                                | 2025         | 2025-11-18   | Not disclosed                     | Native Multimodal Reasoning Model            | Gemini 3 Pro/Flash generation; upgraded reasoning, multimodal understanding, coding, and agentic capabilities.              | —                                                                                                    | <a href="https://blog.google/products/gemini/gemini-3/" class="external text" rel="nofollow">Google Blog</a>                                                                                      |
| **Gemini 3.1**                                                                                                              | 2026         | 2026-02-19   | Not disclosed                     | Native Multimodal Reasoning Model            | Updated Gemini 3 generation; stronger core intelligence, improved agentic behavior, and Pro/Flash/Live variants.            | —                                                                                                    | <a href="https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/" class="external text" rel="nofollow">Google Blog</a>                                            |
| **Gemma 4**                                                                                                                 | 2026         | 2026-04-02   | E2B / E4B / 12B / 31B / 26B A4B   | Multimodal Open-weight; Dense / MoE variants | Open-weight agentic model family; up to 256K context, selected audio support, QAT and MTP variants.                         | —                                                                                                    | <a href="https://ai.google.dev/gemma/docs/core" class="external text" rel="nofollow">Google AI for Developers</a>                                                                                 |
| **Gemini 3.5 Flash**                                                                                                        | 2026         | 2026-05-19   | Not disclosed                     | Native Multimodal Agentic Model              | Frontier-level Flash model optimized for coding, agents, long-horizon workflows, and built-in computer use.                 | —                                                                                                    | <a href="https://deepmind.google/models/gemini/flash/" class="external text" rel="nofollow">Google DeepMind</a>                                                                                   |
| **DiffusionGemma**                                                                                                          | 2026         | 2026-06-10   | 26B MoE                           | Text Diffusion / MoE                         | Experimental open text-generation model using diffusion-style block generation instead of standard token-by-token decoding. | —                                                                                                    | <a href="https://blog.google/innovation-and-ai/technology/developers-tools/diffusion-gemma-faster-text-generation/" class="external text" rel="nofollow">Google Blog</a>                          |

Comparison of major Google / Google DeepMind language models

## Gemini

<table class="wikitable sortable">
<colgroup>
<col style="width: 12%" />
<col style="width: 12%" />
<col style="width: 12%" />
<col style="width: 12%" />
<col style="width: 12%" />
<col style="width: 12%" />
<col style="width: 12%" />
<col style="width: 12%" />
</colgroup>
<thead>
<tr class="header">
<th>No.</th>
<th>Date</th>
<th>Model</th>
<th>Note</th>
<th>Source</th>
<th>Developer</th>
<th>arXiv / Technical report</th>
<th>Model Card</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>1</td>
<td>2023-12-06</td>
<td><strong>Gemini 1.0</strong></td>
<td>First Gemini generation; covered Ultra, Pro and Nano variants; native multimodal model family for text, image, audio and video understanding.</td>
<td><a href="https://blog.google/innovation-and-ai/technology/ai/google-gemini-ai/" class="external text" rel="nofollow">Google</a></td>
<td>—</td>
<td><a href="https://arxiv.org/abs/2312.11805" class="external text" rel="nofollow">arXiv:2312.11805</a><br />
<a href="https://storage.googleapis.com/deepmind-media/gemini/gemini_1_report.pdf" class="external text" rel="nofollow">Google DeepMind PDF</a></td>
<td>HTML: —<br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/gemini/gemini_1_report.pdf" class="external text" rel="nofollow">Google DeepMind PDF</a></td>
</tr>
<tr class="even">
<td>2</td>
<td>2024-02-15</td>
<td><strong>Gemini 1.5</strong></td>
<td>Long-context Gemini generation; includes Gemini 1.5 Pro and Gemini 1.5 Flash.</td>
<td><a href="https://blog.google/innovation-and-ai/products/google-gemini-next-generation-model-february-2024/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://docs.cloud.google.com/vertex-ai/generative-ai/docs/learn/model-versions" class="external text" rel="nofollow">Vertex AI model versions</a></td>
<td><a href="https://arxiv.org/abs/2403.05530" class="external text" rel="nofollow">arXiv:2403.05530</a><br />
<a href="https://storage.googleapis.com/deepmind-media/gemini/gemini_v1_5_report.pdf" class="external text" rel="nofollow">Google DeepMind PDF</a></td>
<td>HTML: —<br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/gemini/gemini_v1_5_report.pdf" class="external text" rel="nofollow">Google DeepMind PDF</a></td>
</tr>
<tr class="odd">
<td>3</td>
<td>2024-12-11</td>
<td><strong>Gemini 2.0 Flash</strong></td>
<td>First Gemini 2.0 model; introduced stronger agentic capabilities, native tool use and multimodal output. GA followed on 2025-02-05.</td>
<td><a href="https://blog.google/innovation-and-ai/models-and-research/google-deepmind/google-gemini-ai-update-december-2024/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-2.0-flash" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: —<br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-2-0-Flash-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="even">
<td>4</td>
<td>2025-02-05</td>
<td><strong>Gemini 2.0 Flash-Lite</strong></td>
<td>Cost-efficient Gemini 2.0 model; preview released on 2025-02-05 and GA followed on 2025-02-25.</td>
<td><a href="https://developers.googleblog.com/en/gemini-2-family-expands/" class="external text" rel="nofollow">Google Developers Blog</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-2.0-flash-lite" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: —<br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-2-0-Flash-Lite-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="odd">
<td>5</td>
<td>2025-03-25</td>
<td><strong>Gemini 2.5 Pro</strong></td>
<td>First Gemini 2.5 “thinking” Pro model; stable GA version followed on 2025-06-17.</td>
<td><a href="https://blog.google/innovation-and-ai/models-and-research/google-deepmind/gemini-model-thinking-updates-march-2025/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-2.5-pro" class="external text" rel="nofollow">Gemini API docs</a></td>
<td><a href="https://arxiv.org/abs/2507.06261" class="external text" rel="nofollow">arXiv:2507.06261</a><br />
<a href="https://storage.googleapis.com/deepmind-media/gemini/gemini_v2_5_report.pdf" class="external text" rel="nofollow">Google DeepMind technical report PDF</a></td>
<td>HTML: —<br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-2-5-Pro-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="even">
<td>6</td>
<td>2025-04-17 / 2025-08-26</td>
<td><strong>Gemini 2.5 Flash / Gemini 2.5 Flash Image</strong></td>
<td>Hybrid reasoning Flash model; the image generation/editing variant was later released as Gemini 2.5 Flash Image / Nano Banana.</td>
<td><a href="https://developers.googleblog.com/en/start-building-with-gemini-25-flash/" class="external text" rel="nofollow">Google Developers Blog</a><br />
<a href="https://developers.googleblog.com/en/introducing-gemini-2-5-flash-image/" class="external text" rel="nofollow">Google Developers Blog</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-2.5-flash" class="external text" rel="nofollow">Gemini API docs</a></td>
<td><a href="https://arxiv.org/abs/2507.06261" class="external text" rel="nofollow">arXiv:2507.06261</a><br />
<a href="https://storage.googleapis.com/deepmind-media/gemini/gemini_v2_5_report.pdf" class="external text" rel="nofollow">Google DeepMind technical report PDF</a></td>
<td>HTML: —<br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-2-5-Flash-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="odd">
<td>7</td>
<td>2025-06-17</td>
<td><strong>Gemini 2.5 Flash-Lite</strong></td>
<td>Fastest and most budget-friendly Gemini 2.5 model; preview on 2025-06-17 and stable GA on 2025-07-22.</td>
<td><a href="https://developers.googleblog.com/en/gemini-2-5-thinking-model-updates/" class="external text" rel="nofollow">Google Developers Blog</a><br />
<a href="https://developers.googleblog.com/en/gemini-25-flash-lite-is-now-stable-and-generally-available/" class="external text" rel="nofollow">Google Developers Blog</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-2.5-flash-lite" class="external text" rel="nofollow">Gemini API docs</a></td>
<td><a href="https://arxiv.org/abs/2507.06261" class="external text" rel="nofollow">arXiv:2507.06261</a><br />
<a href="https://storage.googleapis.com/deepmind-media/gemini/gemini_v2_5_report.pdf" class="external text" rel="nofollow">Google DeepMind technical report PDF</a></td>
<td>HTML: —<br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-2-5-Flash-Lite-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="even">
<td>8</td>
<td>2025-08-01</td>
<td><strong>Gemini 2.5 Deep Think</strong></td>
<td>Enhanced reasoning model/mode for especially complex math, science and coding problems.</td>
<td><a href="https://blog.google/products-and-platforms/products/gemini/gemini-2-5-deep-think/" class="external text" rel="nofollow">Google</a></td>
<td>—</td>
<td><a href="https://arxiv.org/abs/2507.06261" class="external text" rel="nofollow">arXiv:2507.06261</a><br />
<a href="https://storage.googleapis.com/deepmind-media/gemini/gemini_v2_5_report.pdf" class="external text" rel="nofollow">Google DeepMind technical report PDF</a></td>
<td>HTML: —<br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-2-5-Deep-Think-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="odd">
<td>9</td>
<td>2025-10-07</td>
<td><strong>Gemini 2.5 Computer Use</strong></td>
<td>Specialized Gemini 2.5 model for browser/UI-control agents using screenshots and generated UI actions.</td>
<td><a href="https://blog.google/innovation-and-ai/models-and-research/google-deepmind/gemini-computer-use-model/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/computer-use" class="external text" rel="nofollow">Gemini API docs</a></td>
<td><a href="https://arxiv.org/abs/2507.06261" class="external text" rel="nofollow">arXiv:2507.06261</a><br />
<a href="https://storage.googleapis.com/deepmind-media/gemini/gemini_v2_5_report.pdf" class="external text" rel="nofollow">Google DeepMind technical report PDF</a></td>
<td>HTML: —<br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-2-5-Computer-Use-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="even">
<td>10</td>
<td>2025-11-18</td>
<td><strong>Gemini 3 Pro</strong></td>
<td>First Gemini 3 model; major reasoning, multimodal and coding upgrade.</td>
<td><a href="https://blog.google/products-and-platforms/products/gemini/gemini-3/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-3-pro-preview" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: —<br />
PDF: <a href="https://deepmind.google/models/model-cards/gemini-3-pro/" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="odd">
<td>11</td>
<td>2025-11-20</td>
<td><strong>Gemini 3 Pro Image / Nano Banana Pro</strong></td>
<td>Advanced image generation and editing model based on Gemini 3 Pro.</td>
<td><a href="https://blog.google/innovation-and-ai/products/nano-banana-pro/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-3-pro-image" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: —<br />
PDF: <a href="https://deepmind.google/models/model-cards/gemini-3-pro-image/" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="even">
<td>12</td>
<td>2025-12-17</td>
<td><strong>Gemini 3 Flash</strong></td>
<td>Gemini 3 Flash model optimized for frontier-class performance at lower latency and cost.</td>
<td><a href="https://blog.google/products-and-platforms/products/gemini/gemini-3-flash/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-3-flash-preview" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: —<br />
PDF: <a href="https://deepmind.google/models/model-cards/gemini-3-flash/" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="odd">
<td>13</td>
<td>2026-02-19</td>
<td><strong>Gemini 3.1 Pro</strong></td>
<td>Upgraded Gemini 3.1 Pro model for complex reasoning, coding, multimodal tasks and agentic workflows.</td>
<td><a href="https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-3.1-pro-preview" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: <a href="https://deepmind.google/models/model-cards/gemini-3-1-pro/" class="external text" rel="nofollow">Google DeepMind HTML</a><br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-3-1-Pro-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="even">
<td>14</td>
<td>2026-02-26</td>
<td><strong>Gemini 3.1 Flash Image / Nano Banana 2</strong></td>
<td>Gemini 3.1 image generation/editing model based on Gemini 3 Flash.</td>
<td><a href="https://blog.google/innovation-and-ai/products/google-ai-updates-february-2026/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-image" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: <a href="https://deepmind.google/models/model-cards/gemini-3-1-flash-image/" class="external text" rel="nofollow">Google DeepMind HTML</a><br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-3-1-Flash-Image-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="odd">
<td>15</td>
<td>2026-03-03</td>
<td><strong>Gemini 3.1 Flash-Lite</strong></td>
<td>Frontier-class lightweight Gemini 3.1 model focused on cost efficiency and high-frequency tasks.</td>
<td><a href="https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-flash-lite/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: <a href="https://deepmind.google/models/model-cards/gemini-3-1-flash-lite/" class="external text" rel="nofollow">Google DeepMind HTML</a><br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-3-1-Flash-Lite-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="even">
<td>16</td>
<td>2026-03-26 / 2026-04-15</td>
<td><strong>Gemini 3.1 Flash Audio / Flash Live / TTS</strong></td>
<td>Audio-focused Gemini 3.1 Flash card covering Flash Live and text-to-speech variants.</td>
<td><a href="https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-flash-live/" class="external text" rel="nofollow">Google</a><br />
<a href="https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-flash-tts/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: <a href="https://deepmind.google/models/model-cards/gemini-3-1-flash-audio/" class="external text" rel="nofollow">Google DeepMind HTML</a><br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-3-1-Flash-Audio-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="odd">
<td>17</td>
<td>2026-05-19</td>
<td><strong>Gemini 3.5 Flash</strong></td>
<td>Gemini 3.5 Flash model for sustained frontier performance on agentic and coding tasks.</td>
<td><a href="https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-3.5-flash" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: <a href="https://deepmind.google/models/model-cards/gemini-3-5-flash/" class="external text" rel="nofollow">Google DeepMind HTML</a><br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-3-5-Flash-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="even">
<td>18</td>
<td>2026-05-29</td>
<td><strong>Gemini Omni Flash</strong></td>
<td>Conversational multimodal model for video generation and conversational video editing; API public preview followed on 2026-06-30.</td>
<td><a href="https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-omni/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-omni-flash" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: <a href="https://deepmind.google/models/model-cards/gemini-omni-flash/" class="external text" rel="nofollow">Google DeepMind HTML</a><br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-Omni-Flash-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="odd">
<td>19</td>
<td>2026-06-09</td>
<td><strong>Gemini 3.5 Audio / Live Translate</strong></td>
<td>Gemini 3.5 audio model for near real-time speech-to-speech translation.</td>
<td><a href="https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-live-3-5-translate/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: <a href="https://deepmind.google/models/model-cards/gemini-3-5-audio/" class="external text" rel="nofollow">Google DeepMind HTML</a><br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-3-5-Audio-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
<tr class="even">
<td>20</td>
<td>2026-06-30</td>
<td><strong>Gemini 3.1 Flash-Lite Image / Nano Banana Lite</strong></td>
<td>Low-latency, cost-efficient Gemini image generation and editing model for high-volume use cases.</td>
<td><a href="https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-omni-flash-nano-banana-2-lite/" class="external text" rel="nofollow">Google</a></td>
<td><a href="https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite-image" class="external text" rel="nofollow">Gemini API docs</a></td>
<td>—</td>
<td>HTML: <a href="https://deepmind.google/models/model-cards/gemini-3-1-flash-lite-image/" class="external text" rel="nofollow">Google DeepMind HTML</a><br />
PDF: <a href="https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-3-1-Flash-Lite-Image-Model-Card.pdf" class="external text" rel="nofollow">Model Card PDF</a></td>
</tr>
</tbody>
</table>

## Application in products and the ecosystem

Google actively integrates its LLMs across its entire product line:

- **Google Search**: BERT, MUM, and Gemini are used to better understand complex queries and provide direct answers in the form of **AI Overviews** (formerly SGE).
- **Google Assistant and Bard (now Gemini)**: The transition from simple voice commands to full-fledged conversational assistants based on LaMDA, PaLM 2, and Gemini.
- **Google Workspace**: The **Duet AI** features (now Gemini for Workspace) help write emails in Gmail, create texts in Docs, and generate presentations in Slides.
- **Android**: Gemini Nano enables AI features to run locally on devices like Pixel, enhancing privacy and speed.
- **Google Cloud AI**: The **Vertex AI** platform provides businesses with access to PaLM and Gemini models via API to build their own applications.

## External links

- <a href="https://ai.google/" class="external text" rel="nofollow">Official Google AI portal</a>
- <a href="https://deepmind.google/" class="external text" rel="nofollow">Official Google DeepMind website</a>

## Bibliography

- Vaswani, A. et al. (2017). *Attention Is All You Need*. NIPS.
- Devlin, J. et al. (2019). *BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding*. NAACL.
- Raffel, C. et al. (2020). *Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer*. JMLR.
- Thoppilan, R. et al. (2022). *LaMDA: Language Models for Dialog Applications*. arXiv:2201.08239.
- Hoffmann, R. et al. (2022). *Training Compute-Optimal Large Language Models*. arXiv:2203.15556.
- Chowdhery, A. et al. (2022). *PaLM: Scaling Language Modeling with Pathways*. JMLR.
- Gemini Team, Google (2023). *Gemini: A Family of Highly Capable Multimodal Models*. arXiv:2312.11805.