Google

Gemini 3.8 Live Extended Thinking

Google

Gemini 3.8 Live Extended Thinking is the separately named high-reasoning audio-to-audio tier for live interactions. It spends more computation on background reasoning while handling text, image, audio, and video input and generating text and audio. Asynchronous function calls allow tool-driven work during a session. Integrations must monitor interaction status because completion of a response turn does not necessarily end background processing.

Gemini 3.8 Live

Google

Gemini 3.8 Live is a generally available real-time audio-to-audio model for low-latency voice agents and dialogue. Text, image, audio, and video input can be combined in a live session, with text and audio responses. Interleaved reasoning and asynchronous function calling help agents continue an interaction while using tools. It is the standard Live tier; extended background reasoning has a separately named model identity.

Gemini 3.8 Flash

Google

Gemini 3.8 Flash is the stable September update to the Flash line, improving sustained coding, instruction following, and reliable multi-step agent work. It retains multimodal input and a million-token context with text output. Native tools cover search grounding, code execution, function calling, and preview computer use. Low, medium, and high thinking levels balance response time against reasoning computation.

Gemini 3.7 Flash

Google

Gemini 3.7 Flash is a stable model update with substantial improvements in software engineering, web development, and tool-based workflows. It accepts text, images, audio, and video and generates text, with a million-token input context. Low, medium, and high thinking levels control reasoning computation; native function calls, grounded retrieval, and preview computer use support agent integrations.

Gemini Robotics ER 2

Google

Gemini Robotics ER 2 is a public-preview vision-language model for embodied reasoning, based on Gemini 3.5 Flash. It improves spatial reasoning, multi-step tool use, robot-video moment finding and progress classification, and multi-robot coordination. A streaming endpoint adapts the same release to continuous audio and video using the Live API. The model produces text and high-level tool instructions rather than directly serving as a motor controller.

Gemini 3.5 Flash-Lite

Google

Gemini 3.5 Flash-Lite is a generally available efficiency tier for low-latency automation and subagents. It combines a million-token input context with multimodal understanding, thinking, structured outputs, and native tools. Its positioning emphasizes cost-effective high-volume work rather than the deepest reasoning. Computer use is supported as a preview capability and needs a compatible execution environment.

Gemini 3.6 Flash

Google

Gemini 3.6 Flash is a generally available update emphasizing token efficiency, more concise responses, and stronger coding and agentic planning. It maintains multimodal input and long-context text output while improving the cost profile over 3.5 Flash. Thinking levels and native tools support sustained workflows. Sampling controls such as temperature, top-p, and top-k are deprecated for this generation.

DiffusionGemma

Google

DiffusionGemma is an experimental open-weight text diffusion model built on the Gemma 4 26B A4B backbone. An autoregressive encoder caches context while a bidirectional decoder repeatedly denoises 256-token canvases, producing blocks of text in parallel. It accepts text and visual input and supports thinking and tools. It targets fast small-batch generation on a capable accelerator, trading some task accuracy for throughput relative to autoregressive Gemma 4.

Gemma 4 12B

Google

Gemma 4 12B is a separately released multimodal update using a unified, encoder-free architecture. Image patches and audio waveforms enter through linear projections into the language model instead of dedicated modality encoders. Its approximately 12B dense decoder supports text, vision, audio, thinking, and native tool use with a 256K context. Quantized deployment targets consumer hardware, including laptops with 16GB of memory; actual capacity depends on precision and context.

Gemini 3.5 Flash

Google

Gemini 3.5 Flash is a generally available model update for sustained coding and agentic work. It combines multimodal understanding with a million-token input context, thinking, and tools for code execution and grounded retrieval. Computer use was subsequently exposed as a public preview tool for browser, mobile, and desktop workflows. The Antigravity managed agent uses this model but is a separate application.

Gemini Robotics ER 1.6

Google

Gemini Robotics ER 1.6 is the independently released embodied-reasoning update for interpreting physical spaces and planning robot tasks. It advances instrument reading and spatial and physical reasoning, with multimodal input and text output. Code execution, function calling, and structured responses let it coordinate external robot APIs. It was a public API preview, subsequently retired in August 2026.

Gemma 4

Google

Gemma 4 is an open-weight multimodal family with native thinking and function calling, released under Apache 2.0. The initial E2B and E4B models target constrained devices, while 26B A4B uses sparse experts and 31B is dense. All support vision and video frames; audio is available in the smaller models. Context lengths vary by size, and the independently released encoder-free 12B update has its own page.

Gemini 3.1 Flash Live

Google

Gemini 3.1 Flash Live is the public-preview Gemini 3.1 audio-to-audio model for real-time conversational agents. It accepts text, images, audio, and video and returns text and audio, with reasoning and tools during live sessions. It supports native audio output and search grounding. The current catalog describes it as a legacy preview and recommends migration to Gemini 3.8 Live.

Gemini 3.1 Flash-Lite

Google

Gemini 3.1 Flash-Lite is the lower-cost Gemini 3.1 tier for high-frequency, lightweight workloads. Its public preview began in March, followed by general availability in May. It accepts text, image, audio, and video inputs and supports thinking, function calling, and structured responses. It is intended for repeated extraction, summarization, translation, and classification jobs with precise instructions.

Gemini 3.1 Pro

Google

Gemini 3.1 Pro is an independently released Pro update for complex reasoning, software engineering, and multimodal work. It is available as a public API preview with a million-token input context and adjustable thinking. A separate custom-tools endpoint prioritizes developer-supplied tools when built-in and custom tools are combined; that serving variant is represented on this page.

MedGemma 1.5

Google

MedGemma 1.5 is the independently released 4B instruction-tuned update to MedGemma for medical text and image interpretation. It adds CT and MRI volume understanding, whole-slide pathology inputs, longitudinal chest X-ray analysis, anatomical localization, and structured extraction from medical reports and EHRs. It uses Gemma 3 with a medical-data-trained SigLIP image encoder and produces text. The weights are intended as a developer foundation requiring application-specific validation.

FunctionGemma

Google

FunctionGemma is a specialized version of Gemma 3 270M fine-tuned for function calling and designed to run on edge devices. It bridges natural language and software execution, translating user commands into executable API actions. The model excels at unified action and chat capabilities, switching seamlessly between generating structured function calls and conversational responses. Built specifically for customization through fine-tuning, it demonstrated 85% accuracy on Mobile Actions after training (up from 58% baseline). Small enough to run on mobile phones and edge devices like NVIDIA Jetson Nano, it uses Gemma's 256k vocabulary to efficiently tokenize JSON and multilingual inputs.

Gemini 3 Flash

Google

Gemini 3 Flash is the first Flash tier in the Gemini 3 generation, combining multimodal understanding and reasoning with lower latency and cost than Pro. It accepts text, images, audio, and video and produces text, with configurable thinking and native tools. It launched as a public preview through the Gemini API and Google AI Studio and as the default model in the Gemini app.

Gemini 3 Pro

Google

Gemini 3 Pro is Google’s November 2025 reasoning and multimodal model for software engineering, long documents, and tool-based workflows. It accepts text, images, audio, and video and produces text with a million-token input limit. Thinking levels adjust ordinary reasoning effort; Gemini 3 Deep Think is a separately exposed enhanced reasoning mode, first publicly available in December and substantially upgraded in February 2026. The Pro API preview was shut down March 9, 2026, while the Deep Think mode has separate product access.

Gemini 2.5 Computer Use

Google

Gemini 2.5 Computer Use is the Gemini 2.5 specialization trained to interpret screenshots and produce actions for graphical interfaces. It launched as a public API preview for browser automation, including clicking, typing, scrolling, and navigation. Developers execute proposed actions in their own environment and return the resulting screenshots. Its preview endpoint was shut down in July 2026; this page preserves the historical specialization.

Gemini Robotics ER 1.5

Google

Gemini Robotics ER 1.5 is the publicly released vision-language model for embodied reasoning and high-level robot orchestration. It interprets visual scenes, reasons about spatial relationships, creates multi-step plans, and calls developer tools. It was offered through the Gemini API and AI Studio, distinct from the restricted vision-language-action Robotics 1.5 motor model. Its endpoint was retired in April 2026.

Gemini 2.5 Flash-Lite

Google

Google's most cost-efficient and fastest 2.5 model yet. Higher quality than 2.0 Flash-Lite on coding, math, science, reasoning and multimodal benchmarks. Excels at high-volume, latency-sensitive tasks like translation and classification.

Gemini 2.5 Flash

Google

Improved across key benchmarks for reasoning, multimodality, code and long context while getting even more efficient. Best for fast performance on complex tasks.

Gemini 2.5 Flash Live

Google

Gemini 2.5 Flash Live is the native-audio Gemini 2.5 specialization for bidirectional voice and video conversations. The initial native-audio dialog models launched publicly in May 2025; September and December endpoints are subsequent variants. Native audio reasoning, interruptions, tool calls, and search grounding enable interactive assistants. The current preview endpoint is retained for existing users, while new projects are directed to later Live models.

Gemini Diffusion

Google

Google's state-of-the-art, experimental text diffusion model that uses diffusion techniques to explore a new kind of language model that gives users greater control, creativity, and speed in text generation. Unlike traditional autoregressive models that generate text sequentially, diffusion models refine noise step-by-step to generate entire blocks of tokens at once.

Gemma 3n

Google

Gemma 3n is a mobile-first AI model built on a groundbreaking architecture optimized for on-device performance. It features Per-Layer Embeddings (PLE) for reduced RAM usage, allowing larger models to run on mobile devices with a memory footprint comparable to much smaller models. Supports multimodal understanding including text, images, audio, and video.

MedGemma

Google

MedGemma is Google’s open-weight medical language and vision family, based on Gemma 3. The May 2025 release included 4B multimodal and 27B text instruction-tuned models; a 27B multimodal model was added July 9. Medical visual checkpoints use a SigLIP encoder trained on de-identified clinical images. The multimodal 27B variant also targets medical records and FHIR data. These are developer foundations requiring adaptation and validation; the independently released MedGemma 1.5 update has its own page.

Gemini 2.5 Pro

Google

Gemini 2.5 Pro is capable of reasoning through its thoughts before responding, resulting in enhanced performance and improved accuracy. Features Deep Think, an enhanced reasoning mode, and native audio outputs that capture subtle nuances of speech.

TxGemma

Google

TxGemma is an open-weight Gemma 2 specialization fine-tuned for therapeutic development. It uses natural language, chemical SMILES, protein and nucleic-acid sequences to predict properties of therapeutic entities and targets. Prediction checkpoints prioritize task accuracy; the 9B and 27B chat variants add explanations and multi-turn dialogue. The family is a research and development foundation under Health AI Developer Foundations terms.

Gemma 3

Google

Gemma 3 is a collection of lightweight, state-of-the-art open models built from the same research and technology that powers Gemini 2.0 models. The most capable model you can run on a single GPU or TPU, delivering state-of-the-art performance for its size and outperforming Llama3-405B, DeepSeek-V3 and o3-mini in preliminary human preference evaluations on LMArena's leaderboard. Features advanced text and visual reasoning capabilities, 128K token context window, function calling, and support for over 140 languages.

Gemini 2.0 Flash-Lite

Google

Best for cost-efficient performance. Better quality than 1.5 Flash, at the same speed and cost. Features a 1 million token context window and multimodal input.

Gemini 2.0 Flash

Google

Next iteration of Gemini, released in three versions (Flash, Flash-Lite, Pro). Represents Google's state-of-the-art multimodal model, likely incorporating Mixture-of-Experts at unprecedented scale and targeting dominance in both text and image understanding.

Gemini 2.0 Pro

Google

Google's best model yet for coding performance and complex prompts, with better understanding and reasoning of world knowledge than any previous release. Features a massive 2 million token context window and the ability to call tools like Google Search and code execution.

PaliGemma 2

Google

PaliGemma 2 is the next-generation open-weight vision-language family combining the SigLIP image encoder with Gemma 2 decoders. Three model sizes and three image resolutions offer choices for adaptation and compute constraints. It is designed for fine-tuning on visual tasks including OCR, image captioning, document understanding, localization, and question answering. Later mix fine-tunes are variants of this release rather than separate model generations.

DataGemma

Google

DataGemma is an open-weight Gemma 2 specialization that connects language generation to public statistics in Google Data Commons. Retrieval-interleaved generation annotates statistics with queries for verification, while retrieval-augmented generation prepares questions for external retrieval and synthesis. The released RIG and RAG checkpoints are components of a complete retrieval workflow. They are early research models intended to study grounding rather than standalone general-purpose assistants.

Gemma 2

Google

Gemma 2 is the second open-weight Gemma generation, initially released at 9B and 27B, with a 2B tier added in July. Its architecture alternates local and global attention and applies soft capping for training stability. The smaller tiers use knowledge distillation. Pre-trained and instruction-tuned weights support local text generation, reasoning, and adaptation under the custom Gemma license.

PaliGemma

Google

PaliGemma is an open-weight vision-language model intended for task-specific fine-tuning. It combines a SigLIP visual encoder and Gemma text decoder, accepting images and text and generating text or structured tokens. Applications include captioning, visual question answering, OCR, detection, and segmentation. Pre-trained checkpoints at multiple image resolutions and task-mixture fine-tunes share this catalog page; it is not positioned as a general multi-turn chat assistant.

CodeGemma

Google

CodeGemma is the open-weight Gemma specialization for code generation and completion. The 2B model targets fast completion, while 7B checkpoints cover code-focused pretraining and instruction-tuned assistance. Fill-in-the-middle training enables inserting code between surrounding context, complementing ordinary completion. Training covers programming languages, natural-language discussion of code, and mathematical data, with deployment governed by the Gemma license.

RecurrentGemma

Google

RecurrentGemma is Google’s open-weight language model family based on the Griffin hybrid architecture. Gated linear recurrences and local attention reduce the growing memory demands of ordinary attention-based decoding, benefiting long generated sequences and larger batches. Pre-trained and instruction-tuned checkpoints support text generation, summarization, reasoning, and language-model research. The model family uses the custom Gemma terms.

Gemma

Google

Gemma is Google’s original open-weight language model family, released as 2B and 7B pre-trained and instruction-tuned checkpoints. It builds on research used for Gemini while targeting accessible local deployment. The dense decoder models generate text and can be adapted for question answering, summarization, reasoning, and coding. The custom Gemma terms govern distribution and use; open weights do not imply release of the complete training data.

Gemini 1.5

Google

Upgrade over Gemini 1.0, featuring a **1 million+ token context window** (initially 1M, later expanded to 2M). Likely uses a Mixture-of-Experts backend and improved multimodal abilities. Released in variants (e.g. Flash).

Gemini 1.0

Google

First version of Google's Gemini, a multimodal model built by the joint Google DeepMind team. Comes in three sizes; integrated into a chatbot (Google's next-gen model after PaLM 2) with planned image and text capabilities.

PaLM 2

Google

Second-generation PaLM, actually a family of models (up to ~340B dense params) with training on a diverse multilingual dataset. Powers Google's Bard and other products; improved coding, reasoning, and multilingual skills over PaLM.

Med-PaLM 2

Google

Med-PaLM 2 is the second published medical-language research model, introduced at Google’s March 2023 health event and subsequently documented in a research paper. Built on PaLM 2 with medical-domain training, it improves exam question answering and physician-rated long-form responses. The model informed the later MedLM product family, whose deployment identities are separate. Public weights were not released, and research evaluation alone does not establish clinical readiness.

Flan-UL2

Google

Flan-UL2 is the independently released instruction-tuned UL2 update. Continued training extends the receptive field to 2048 tokens and removes the requirement for mode-switch tokens before Flan instruction tuning. The 20B T5-style encoder-decoder supports text-to-text generation, reasoning, and few-shot prompting. Google’s UL2 repository records checkpoint release on February 28, 2023, before later model-card and ecosystem announcements.

Med-PaLM

Google

Med-PaLM is Google’s published medical-language research model using instruction prompt tuning to adapt Flan-PaLM to medical question answering. The study evaluates medical exams and long-form consumer health questions, including physician assessment of answer quality and harm. It demonstrated medical-domain adaptation of a large language model but did not release weights or a general inference endpoint. Its publication qualifies as a historical research milestone.

Flan-T5

Google

Flan-T5 is the instruction-tuned T5 family released with research on scaling instruction fine-tuning. It mixes a large collection of prompted NLP tasks and chain-of-thought examples to improve zero-shot and few-shot generalization. Checkpoints from Small to XXL preserve the encoder-decoder design and support text-to-text tasks. It is a significant instruction-specialized family, separate from the original T5 pretraining release.

UL2

Google

UL2 is a research and open-weight text-to-text language model using a mixture-of-denoisers pretraining objective. It combines regular span corruption, sequential denoising, and extreme corruption to unify language-understanding and generation tasks. The released 20B model uses a T5-style encoder-decoder and mode tokens that select the desired learning behavior. It became a foundation for instruction tuning and chain-of-thought research.

PaLM

Google

540B-parameter Transformer model trained with the Pathways system. Showed strong few-shot performance and was the largest dense model publicly disclosed as of 2022.

Chinchilla

Google

70B-parameter model trained on 4× more data than typical, to be compute-optimal. Achieved superior accuracy vs much larger models, validating data-scaling laws.

LaMDA

Google

137B-parameter dialogue-optimized language model ("Language Model for Dialogue Applications"). Known for conversational ability and for sparking discussions on AI sentience.

GLaM (Generalist Language Model)

Google

A 1.2 trillion-parameter **Mixture-of-Experts** Transformer model. Uses sparse activation (64 experts) to reduce inference cost relative to model size.

Gopher

Google

A 280B-parameter Transformer model from DeepMind, used to explore scaling laws and as a precursor to Chinchilla.

T5 (Text-To-Text Transfer Transformer)

Google

Encoder-decoder Transformer trained on a unified text-to-text format for diverse NLP tasks. Enabled prompt-based multitask learning.

XLNet

Google

Autoregressive Transformer that uses permuted language modeling as an alternative to BERT. Achieved state-of-the-art on several NLP tasks in 2019.

BERT

Google

Bidirectional Transformer pre-trained for language understanding (not generative). Influential for NLP tasks with masked-language modeling.