> ## AI Agent Context > Get the complete documentation index at: https://aistudio.google.com/llms.txt > Use this file to discover all available pages. > For the full documentation in a single file, fetch: https://aistudio.google.com/llms-full.txt # Latest Gemini models Build with Google's latest models for reasoning, coding, and native media generation through one unified API. Start with `gemini-4-hydrogen` for fast, dependable coding, agentic workflows, and multimodal knowledge work. Use `gemini-4-carbon` or `gemini-4-argon` for deep frontier reasoning, or pick a specialized model below. {% button href="/docs/thinking" variant="ais" %}Choosing a model{% /button %} {% button href="/docs/models" variant="primary" %}View all models{% /button %} ## Compare Gemini models {% class="hide-from-toc" id="compare-models" %} Every Gemini model below accepts text, images, audio, video, and PDFs. All four models support [Google Search grounding](/docs/google-search), [code execution](/docs/code-execution), [function calling](/docs/function-calling), [computer use](/docs/computer-use), and [structured outputs](/docs/structured-output). {% card-grid columns="4" %} {% card title="Gemini 4 Carbon" variant="model-card" %} {% section class="model-specs" %} * *Model ID* **`gemini-4-carbon`** * *Price (1M)* **$2.00 / $10.00 (intro thru Nov 30)** * *Speed* **Deliberate** * *Context* **1M unified tokens** * *Max output* **64K default / 1M [long decode](/docs/long-decode)** * *Thinking* **`low`, `medium` (default), `high`** {% /section %} {% section class="card-actions" %} {% button href="https://aistudio.google.com/prompts/new_chat?model=gemini-4-carbon" tooltip="Try Gemini 4 Carbon in AI Studio" variant="primary" %}Try in AI Studio{% /button %} {% button href="/docs/models/gemini-4-carbon" icon="arrow_forward" tooltip="View Gemini 4 Carbon model specs" variant="icon-primary" /%} {% /section %} {% /card %} {% card title="Gemini 4 Argon" variant="model-card" %} {% section class="model-specs" %} * *Model ID* **`gemini-4-argon`** * *Price (1M)* **$2.00 / $10.00 (intro thru Nov 30)** * *Speed* **Moderate** * *Context* **1M unified tokens** * *Max output* **64K default / 1M [long decode](/docs/long-decode)** * *Thinking* **`low`, `medium` (default), `high`** {% /section %} {% section class="card-actions" %} {% button href="https://aistudio.google.com/prompts/new_chat?model=gemini-4-argon" tooltip="Try Gemini 4 Argon in AI Studio" variant="primary" %}Try in AI Studio{% /button %} {% button href="/docs/models/gemini-4-argon" icon="arrow_forward" tooltip="View Gemini 4 Argon model specs" variant="icon-primary" /%} {% /section %} {% /card %} {% card title="Gemini 4 Hydrogen" variant="model-card" %} {% section class="model-specs" %} * *Model ID* **`gemini-4-hydrogen`** * *Price (1M)* **$1.50 / $7.50** * *Speed* **Fast** * *Context* **1M unified tokens** * *Max output* **64K default / 1M [long decode](/docs/long-decode)** * *Thinking* **`low`, `medium`, `high` (default), `xhigh`** {% /section %} {% section class="card-actions" %} {% button href="https://aistudio.google.com/prompts/new_chat?model=gemini-4-hydrogen" tooltip="Try Gemini 4 Hydrogen in AI Studio" variant="primary" %}Try in AI Studio{% /button %} {% button href="/docs/models/gemini-4-hydrogen" icon="arrow_forward" tooltip="View Gemini 4 Hydrogen model specs" variant="icon-primary" /%} {% /section %} {% /card %} {% card title="Gemini 3.5 Flash-Lite" variant="model-card" %} {% section class="model-specs" %} * *Model ID* **`gemini-3.5-flash-lite`** * *Price (1M)* **$0.30 / $2.50** * *Speed* **Fastest** * *Context* **1M input tokens** * *Max output* **64K tokens** * *Thinking* **`minimal` (default), `low`, `medium`, `high`** {% /section %} {% section class="card-actions" %} {% button href="https://aistudio.google.com/prompts/new_chat?model=gemini-3.5-flash-lite" tooltip="Try Gemini 3.5 Flash-Lite in AI Studio" variant="primary" %}Try in AI Studio{% /button %} {% button href="/docs/models/gemini-3.5-flash-lite" icon="arrow_forward" tooltip="View Gemini 3.5 Flash-Lite model specs" variant="icon-primary" /%} {% /section %} {% /card %} {% /card-grid %} ## What's new guides {% class="hide-from-toc" id="whats-new-guides" %} Read the What's new guides for new features, API changes, code examples, and migration steps. {% card-grid columns="4" %} {% card description="Our strongest reasoning model for agentic coding, multimodal precision, scientific and security reasoning, and choosing between Carbon and Argon." href="/docs/whats-new-gemini-4-carbon" icon="spark" title="What's new in Gemini 4 Carbon" variant="stacked-icon" /%} {% card description="Our efficient Gemini 4 frontier reasoning model, with faster responses and lower token spend per turn compared to Carbon, unified 1M context, and long decode." href="/docs/whats-new-gemini-4-argon" icon="psychology" title="What's new in Gemini 4 Argon" variant="stacked-icon" /%} {% card description="Cutting-edge intelligence with speed and efficiency for everyday coding and agentic workflows, featuring four thinking levels up to xhigh." href="/docs/whats-new-gemini-4-hydrogen" icon="spark" title="What's new in Gemini 4 Hydrogen" variant="stacked-icon" /%} {% card description="Fastest, lowest-cost Flash-Lite model for high-throughput subagents, routing, and step-by-step migration from Gemini 2.5." href="/docs/whats-new-gemini-3.6" icon="bolt" title="What's new in Gemini 3.6 and 3.5 Flash-Lite" variant="stacked-icon" /%} {% /card-grid %} ## Latest models across modalities {% class="hide-from-toc" id="latest-by-modality" %} Pick the current recommended model ID when starting a project across text, images, video, or speech. {% card-grid columns="2" %} {% card description="Multimodal understanding, software engineering, structured extraction, and tool-using agents." icon="psychology" title="Text, code, and reasoning" variant="stacked-icon" %} {% section class="model-specs" %} * *Flagship reasoning* **[`gemini-4-carbon`](/docs/models/gemini-4-carbon), [`gemini-4-argon`](/docs/models/gemini-4-argon)** * *Default general-purpose* **[`gemini-4-hydrogen`](/docs/models/gemini-4-hydrogen)** * *High-volume and low-cost* **[`gemini-3.5-flash-lite`](/docs/models/gemini-3.5-flash-lite)** {% /section %} {% section class="card-actions" %} {% button href="https://aistudio.google.com/prompts/new_chat?model=gemini-4-hydrogen" tooltip="Open text chat in AI Studio" variant="primary" %}Try in AI Studio{% /button %} {% button href="/docs/text-generation" icon="arrow_forward" tooltip="Read text generation guide" variant="icon-primary" /%} {% /section %} {% /card %} {% card description="Conversational image generation, multi-turn editing, character consistency, and accurate text rendering up to 4K." icon="banana_spark" title=Images (Nano Banana) variant="stacked-icon" %} {% section class="model-specs" %} * *Default high-efficiency* **[`gemini-nano-banana-2.1`](/docs/models/gemini-nano-banana-2.1)** * *Studio 4K and layouts* **[`gemini-3-pro-image`](/docs/models/gemini-3-pro-image)** * *Ultra-low latency* **[`gemini-3.1-flash-lite-image`](/docs/models/gemini-3.1-flash-lite-image)** {% /section %} {% section class="card-actions" %} {% button href="https://aistudio.google.com/prompts/new_chat?model=gemini-nano-banana-2.1" tooltip="Try Nano Banana 2.1 in AI Studio" variant="primary" %}Try in AI Studio{% /button %} {% button href="/docs/image-generation" icon="arrow_forward" tooltip="Read image generation guide" variant="icon-primary" /%} {% /section %} {% /card %} {% card description="Generate and edit up to 4K video with synchronized audio using Gemini Omni, stream live video, or analyze long videos." icon="movie_filter" title="Video" variant="stacked-icon" %} {% section class="model-specs" %} * *Video generation and editing* **[`gemini-omni-1.2-flash`](/docs/models/gemini-omni-flash)** * *Video understanding* **[`gemini-4-carbon`](/docs/models/gemini-4-carbon), [`gemini-4-argon`](/docs/models/gemini-4-argon), [`gemini-4-hydrogen`](/docs/models/gemini-4-hydrogen)** * *Live video streaming* **[`gemini-3.8-live`](/docs/models/gemini-3.8-live)** {% /section %} {% section class="card-actions" %} {% button href="https://aistudio.google.com/prompts/new_chat?model=gemini-omni-1.2-flash" tooltip="Try Gemini Omni 1.2 Flash in AI Studio" variant="primary" %}Try in AI Studio{% /button %} {% button href="/docs/omni" icon="arrow_forward" tooltip="Read Gemini Omni Flash guide" variant="icon-primary" /%} {% /section %} {% /card %} {% card description="Sub-second bidirectional voice agents, expressive text-to-speech in 130 languages, and diarized transcription." icon="graphic_eq" title="Audio and speech" variant="stacked-icon" %} {% section class="model-specs" %} * *Live voice agents* **[`gemini-3.8-live`](/docs/models/gemini-3.8-live)** * *Studio text-to-speech* **[`gemini-3.8-flash-tts`](/docs/models/gemini-3.8-flash-tts)** * *Speech-to-text* **[`gemini-3.5-transcribe`](/docs/models/gemini-3.5-transcribe)** {% /section %} {% section class="card-actions" %} {% button href="https://aistudio.google.com/live" tooltip="Try Gemini 3.8 Live in AI Studio" variant="primary" %}Try in AI Studio{% /button %} {% button href="/docs/live-api" icon="arrow_forward" tooltip="Read Live API guide" variant="icon-primary" /%} {% /section %} {% /card %} {% /card-grid %} ## Quickstart {% class="hide-from-toc" id="quickstart" %} Send your first request using the Google GenAI SDK and the [Interactions API](/docs/interactions-overview). Swap `model` to test any tier. {% tabs %} {% tab filename="quickstart.py" label="Python" %} ```python from google import genai client = genai.Client() interaction = client.interactions.create( model="gemini-4-carbon", input="Explain gravity to a skeptical cat in three short sentences.", ) print(interaction.output_text) ``` {% /tab %} {% tab filename="quickstart.js" label="JavaScript" %} ```javascript import { GoogleGenAI } from "@google/genai"; const client = new GoogleGenAI({}); const interaction = await client.interactions.create({ model: "gemini-4-carbon", input: "Explain gravity to a skeptical cat in three short sentences.", }); console.log(interaction.output_text); ``` {% /tab %} {% tab label="REST" %} ```bash curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions" \ -H "x-goog-api-key: $GEMINI_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "gemini-4-carbon", "input": "Explain gravity to a skeptical cat in three short sentences." }' ``` {% /tab %} {% /tabs %} ## Set up your coding agent {% class="hide-from-toc" id="coding-agents" %} Give your coding agent live access to the latest Gemini models, SDK syntax, and documentation so it always builds with up-to-date information. ```bash # Connect the live Gemini Docs MCP server npx add-mcp "https://gemini-api-docs-mcp.dev" # Install official Gemini API agent skills npx skills add google-gemini/gemini-skills --skill gemini-api-dev --global # Or install the bundled Antigravity plugin agy plugin install https://github.com/google-gemini/gemini-skills ``` You can also append `.md` to any documentation URL or feed [`llms.txt`](https://aistudio.google.com/llms.txt) to your agent. See [Coding agent setup](/docs/coding-agents) for editor configs and `AGENTS.md` templates. ## Specialized models {% class="hide-from-toc" id="specialized-models" %} Beyond the general-purpose and media models above, the Gemini API includes specialized endpoints: * **[Managed Agents](/docs/agents):** [`antigravity-preview-09-2026`](/docs/antigravity-agent) for coding and tool orchestration, and [`deep-research-preview-04-2026`](/docs/deep-research) for autonomous multi-step research. * **[Decisions](/docs/decisions-api):** [`gemini-decision-1`](/docs/models/gemini-decision-1) for fast, calibrated multimodal yes/no, choice, and rubric score decisions with zero output tokens. * **[Embeddings](/docs/embeddings):** [`gemini-embedding-2`](/docs/models/gemini-embedding-2) for multimodal text, image, audio, video, and PDF search and RAG. * **[Music generation](/docs/music-generation):** [`lyria-3.5`](/docs/models/lyria-3.5) for studio-grade instrumental and vocal tracks. * **[Robotics](/docs/robotics-overview):** [`gemini-robotics-er-2-preview`](/docs/models/gemini-robotics-er-2-preview) for spatial reasoning and embodied planning. ## What's next {% class="hide-from-toc" id="whats-next" %} * Browse the full endpoint catalog on [All models](/docs/models). * Tune reasoning depth and latency in the [Thinking guide](/docs/thinking). * Review [Pricing](/docs/pricing), [Rate limits](/docs/rate-limits), [Context caching](/docs/caching), and [Batch API (50% cost reduction)](/docs/batch-api). * Check sunset dates on [Deprecations](/docs/deprecations) and recent updates in [Release notes](/docs/changelog).