# 4ALL API Docs

> Documentation for 4ALL API, an enterprise LLM API gateway: one API key for GPT, Claude, Gemini and 200+ image/video models, fully OpenAI-compatible.

## 产品基础 / Basics

- [Changelog](https://docs.4allapi.com/en/changelog/): Update records for the 4ALL API platform and models.
- [Three Steps to Get Started](https://docs.4allapi.com/en/getting-started/): Register an account, create an API token, and send your first request — start using 4ALL API in three steps.
- [Basic Introduction](https://docs.4allapi.com/en/): 4ALL API enterprise-grade large model API aggregation gateway — one Key access to GPT, Claude, Gemini, and 200+ models.
- [Platform Updates](https://docs.4allapi.com/en/news/): New model releases and platform feature updates for 4ALL API.
- [Pricing](https://docs.4allapi.com/en/pricing/): 4ALL API billing methods: usage-based billing, per-request billing, credits, and group explanations.

## API

- [Authentication and Keys](https://docs.4allapi.com/en/api/authentication/): 4ALL API authentication methods, token management, and security recommendations.
- [Balance Inquiry API](https://docs.4allapi.com/en/api/balance-query/): Programmatically query quota and usage through OpenAI-compatible billing endpoints.
- [Chat Completions](https://docs.4allapi.com/en/api/chat-completions/): /v1/chat/completions interface reference: request parameters, response structure, and examples.
- [Error Codes and Retries](https://docs.4allapi.com/en/api/errors/): Explanations of common 4ALL API error codes and retry recommendations.
- [Log Query API](https://docs.4allapi.com/en/api/log-query/): Use an API token to directly query the call logs for that token.
- [Multi-turn Conversations](https://docs.4allapi.com/en/api/multi-turn/): Use the messages array to maintain context and enable continuous conversations.
- [Streaming Output](https://docs.4allapi.com/en/api/streaming/): Use the stream parameter to receive model output via SSE streaming.
- [Token Management API](https://docs.4allapi.com/en/api/token-management/): Programmatically create, query, modify, and delete API tokens using a system access token.

## Capabilities

- [Abra Video Generation](https://docs.4allapi.com/en/capabilities/abra/): Abra series: text-to-video and reference-image-to-video with 8s/10s durations.
- [Claude Native Format](https://docs.4allapi.com/en/capabilities/claude-native/): Call Claude series models using Anthropic’s native protocol via /v1/messages.
- [Claude Cache Billing](https://docs.4allapi.com/en/capabilities/claude-prompt-caching/): Explicitly cache long prefixes with cache_control to significantly reduce Claude repeated input costs.
- [Streaming and Non-Streaming Responses](https://docs.4allapi.com/en/capabilities/claude-response-handling/): Response structure and streaming event parsing for Claude native Messages API.
- [Thinking Mode](https://docs.4allapi.com/en/capabilities/claude-thinking/): Usage and output parsing of Claude -thinking tier and native thinking parameters.
- [Deprecated Models List](https://docs.4allapi.com/en/capabilities/deprecated-models/): A list of models that have been taken offline or are planned to be taken offline, along with migration recommendations.
- [Text Embeddings](https://docs.4allapi.com/en/capabilities/embeddings/): /v1/embeddings API: converts text into vectors for retrieval and RAG.
- [Free Models](https://docs.4allapi.com/en/capabilities/free-models/): A list of zero-cost models in the free group and their usage limits.
- [FC Function Calling](https://docs.4allapi.com/en/capabilities/function-calling/): Let the model call your functions via the tools parameter, with a unified usage pattern for GPT/Claude/Gemini.
- [Gemini Native Format](https://docs.4allapi.com/en/capabilities/gemini-native/): Call using Google Gemini native generateContent protocol.
- [Web Search (Google Search)](https://docs.4allapi.com/en/capabilities/gemini-web-search/): Use -google-search models to let Gemini answer based on real-time search results.
- [Text-to-Image API Reference](https://docs.4allapi.com/en/capabilities/gpt-image-t2i/): Complete reference for GPT-Image series /v1/images/generations request parameters and responses.
- [Dialogue and Reasoning](https://docs.4allapi.com/en/capabilities/grok-chat/): How to use Grok conversation calls and reasoning tiers.
- [Grok Overview](https://docs.4allapi.com/en/capabilities/grok-overview/): Overview of xAI Grok series access methods and capabilities on 4ALL API.
- [Web Search and X Search](https://docs.4allapi.com/en/capabilities/grok-web-search/): Grok’s real-time web retrieval and X(Twitter) ecosystem search capabilities.
- [Image Invocation Notes (Best Practices)](https://docs.4allapi.com/en/capabilities/image-best-practices/): Key points for timeout, retries, aspect ratio, and cost control in the image generation API. Read before calling.
- [Connection Interruption Troubleshooting](https://docs.4allapi.com/en/capabilities/image-connection-drops/): Common causes of image generation request timeouts and disconnections, with step-by-step troubleshooting methods.
- [Image Editing API](https://docs.4allapi.com/en/capabilities/image-edit/): Edit images via /v1/images/edits: upload the original image + prompt to generate edited results.
- [Image Generation (GPT-Image)](https://docs.4allapi.com/en/capabilities/image-generation/): /v1/images/generations and /v1/images/edits endpoints: text-to-image, image-to-image, and resolution variants.
- [How to Generate Satisfying Images](https://docs.4allapi.com/en/capabilities/image-prompt-tips/): Structured prompt writing and iterative techniques for image prompts.
- [Image Compression and Output Resolution](https://docs.4allapi.com/en/capabilities/image-resolution/): Resolution tier selection, output format, and size control.
- [Image and Video Generation Models](https://docs.4allapi.com/en/capabilities/image-video-models/): Overview of 4ALL API’s image/video models currently on sale and the model naming rules.
- [Claude Series](https://docs.4allapi.com/en/capabilities/models-claude/): 4ALL API's list of Claude models for sale, thinking tiers, and selection recommendations.
- [Gemini Series](https://docs.4allapi.com/en/capabilities/models-gemini/): 4ALL API’s current Gemini model lineup: text, web-connected, and multimodal tiers.
- [GPT Series](https://docs.4allapi.com/en/capabilities/models-gpt/): 4ALL API's list of GPT models for sale, tier differences, and selection recommendations.
- [Model Overview](https://docs.4allapi.com/en/capabilities/models/): 4ALL API’s major model families and selection recommendations.
- [Text-to-Image/Edit API Reference](https://docs.4allapi.com/en/capabilities/nano-banana-reference/): Parameter reference for the generation and editing APIs in the Nano Banana (Gemini Images) series.
- [Nano Banana Images](https://docs.4allapi.com/en/capabilities/nano-banana/): Text-to-image and image-to-image usage for the Nano Banana series of image models.
- [Omni Flash Video Generation](https://docs.4allapi.com/en/capabilities/omni-flash/): Fast video model: 4s–10s duration, portrait and landscape aspect ratios, supports text-to-video, image-to-video, and reference-to-video.
- [Compatibility Mode Invocation (SDK Integration)](https://docs.4allapi.com/en/capabilities/openai-compatible/): Seamlessly integrate with 4ALL API using the official OpenAI SDK, changing only the Base URL and API Key.
- [Native Invocation (Responses API)](https://docs.4allapi.com/en/capabilities/openai-native/): How to use OpenAI's new Responses interface on 4ALL API.
- [Reasoning Model Outputs](https://docs.4allapi.com/en/capabilities/openai-reasoning/): Usage and output interpretation of reasoning tier suffixes such as -high/-low/-thinking.
- [Response Data Handling](https://docs.4allapi.com/en/capabilities/openai-responses/): Proper handling of Chat Completions response structure, usage fields, and finish_reason.
- [Cache Billing](https://docs.4allapi.com/en/capabilities/prompt-caching/): How prompt cache hits are identified and how cache multiplier billing works.
- [Text Generation (Chat Completions)](https://docs.4allapi.com/en/capabilities/text-generation/): Overview of text generation capabilities: APIs, model families, and selection paths.
- [VEO 3.1 Video Generation](https://docs.4allapi.com/en/capabilities/veo-3-1/): Google VEO 3.1 full model lineup: text-to-video, image-to-video, reference-to-video, with standard, fast, and lite tiers.
- [Video Generation (Async)](https://docs.4allapi.com/en/capabilities/video-generation/): /v1/video/generations submit-query asynchronous protocol, supports video models such as Veo, Sora, Kling, Omni Flash, etc.
- [Video Understanding API](https://docs.4allapi.com/en/capabilities/video-understanding/): Analyze video content with the Gemini series models: scene recognition, content summarization, and timeline Q&A.
- [Visual Understanding](https://docs.4allapi.com/en/capabilities/vision/): Pass images in Chat Completions to let the model read, recognize, and analyze them.

## Scenarios

- [Bulk Content Production](https://docs.4allapi.com/en/scenarios/batch-content/): Use text, image, and video models to batch-generate short video scripts, cover images, ad creatives, and social media content.
- [Code-Assisted Development](https://docs.4allapi.com/en/scenarios/code-assisted-dev/): Unified access to Claude, GPT, Gemini, and other coding models, matching the most suitable assistant for different tech stacks.
- [E-commerce Visual Marketing](https://docs.4allapi.com/en/scenarios/ecommerce-visual/): Generate product scene images, model try-on images, and marketing posters, reducing shooting and outsourcing costs.
- [Financial Research Report Generation](https://docs.4allapi.com/en/scenarios/financial-research/): Call a long-context model to analyze financial reports and announcements, and generate structured research summaries and reports.
- [Personalized Educational Tutoring](https://docs.4allapi.com/en/scenarios/personalized-education/): Dynamically select models based on student level, covering everything from basic Q&A to competition coaching.
- [Intelligent Customer Support Upgrade](https://docs.4allapi.com/en/scenarios/smart-customer-service/): Automatically assign models based on question complexity, balancing experience, cost, and stability.

## Tools

- [Chatbox](https://docs.4allapi.com/en/tools/chatbox/): Connect 4ALL API in Chatbox.
- [Cherry Studio](https://docs.4allapi.com/en/tools/cherry-studio/): Connect 4ALL API in Cherry Studio.
- [Claude Code](https://docs.4allapi.com/en/tools/claude-code/): Enable Claude Code to work through 4ALL API’s Anthropic native protocol.
- [Cline(VS Code)](https://docs.4allapi.com/en/tools/cline/): Connect 4ALL API in the Cline extension for VS Code.
- [Cursor](https://docs.4allapi.com/en/tools/cursor/): Use 4ALL API models in the Cursor editor.
- [Dify](https://docs.4allapi.com/en/tools/dify/): Connect 4ALL API as the model provider on the Dify platform.
- [Immersive Translation](https://docs.4allapi.com/en/tools/immersive-translate/): Use 4ALL API models in the Immersive Translation extension for web/document translation.
- [LangChain](https://docs.4allapi.com/en/tools/langchain/): Using 4ALL API in LangChain (Python / JS).
- [Open WebUI](https://docs.4allapi.com/en/tools/open-webui/): Connect 4ALL API in Open WebUI.

## FAQ

- [Billing and Quotas](https://docs.4allapi.com/en/faq/billing/): 4ALL API’s billing model, quotas, and usage lookup.
- [Rate Limiting and 429](https://docs.4allapi.com/en/faq/rate-limits/): Reasons for triggering 429 and the correct retry approach.
- [Data Security](https://docs.4allapi.com/en/faq/security/): 4ALL API’s handling of request data and compliance recommendations.
- [Service Availability](https://docs.4allapi.com/en/faq/sla/): 4ALL API’s high availability architecture and failure handling mechanisms.

## News

- [llm-mcp-client](https://docs.4allapi.com/en/news/2026-07-31-llm-mcp-client-0-1a0/): llm-mcp-client 0.1a0 released: an
- [deepseek-ai/DeepSeek-](https://docs.4allapi.com/en/news/2026-07-31-deepseek-ai-deepseek-v4-flash-0731/): DeepSeek-V4-Flash-
- [Building abundant](https://docs.4allapi.com/en/news/2026-07-31-building-abundant-intelligence/): A full-stack approach to making advanced AI more capable
- [[AINews] GPT 5.6 price cut by 20%-80%: Cost of GPT 5.4 Intelligence dropped 13x in 4 months due to GPT 5.6 recursive self-optimization](https://docs.4allapi.com/en/news/2026-07-31-ainews-gpt-5-6-price-cut-by-20-80-cost-of-gpt-5-4-intelligence-dropped-13x-in-4-/): TL;DR: Knowledge distillation is the key
- [前沿`](https://docs.4allapi.com/en/news/2026-07-30-通过-gpt-5-6-推动价格-性能前沿发展/): on Luna and Terra offers lower pricing,
- [Pushing the Price](https://docs.4allapi.com/en/news/2026-07-30-用-gpt-5-6-推动价格性能边界的提升/): GPT-5.6 price cuts on Luna and Terra enable cheaper, faster AI
- [Ontologies](https://docs.4allapi.com/en/news/2026-07-30-ontologies-are-so-back-why-ai-agents-are-reviving-the-semantic-web/): AI engineers are rediscovering ontologies as a way to keep probabilistic agents inside
- [GPU Management: Why Idle GPUs](https://docs.4allapi.com/en/news/2026-07-30-gpu-管理-为何闲置-gpu-成了新的停飞飞机/): Dharma-AI''s Hugging Face blog covers their latest AI models,
- [Gemini](https://docs.4allapi.com/en/news/2026-07-30-gemini-robotics-er-2-powering-robotics-with-video-understanding-task-orchestrati/): Gemini Robotics ER 2 advances robot reasoning, video understanding,
- [How avatarin Builds](https://docs.4allapi.com/en/news/2026-07-30-avatarin-如何借助-gpt-realtime-构建-24-7-零售智能体/): 23456789012345
- [`: to build / to create](https://docs.4allapi.com/en/news/2026-07-30-avatarin-如何借助-gpt-realtime-打造-24-7-全天候零售代理/): Wait, let''s write a full sentence and
- [Accelerating Scientific Discovery for](https://docs.4allapi.com/en/news/2026-07-29-用-chatgpt-助力学术研究者加速科学发现/): OpenAI grants 100,000 researchers free
- [Tripling Our Score on the](https://docs.4allapi.com/en/news/2026-07-29-启用两个设置后-我们在-arc-agi-3-基准上的得分提高了三倍/): Retaining reasoning and enabling compression boosts GPT-5.6''s ARC
- [Enabling Two Settings Tripled Our](https://docs.4allapi.com/en/news/2026-07-29-启用两个设置后-我们在-arc-agi-3-基准上的得分提升了三倍/): Preserving reasoning and enabling compaction via two API settings boosts GPT-5
- [Kimi K3 Just Broke](https://docs.4allapi.com/en/news/2026-07-29-kimi-k3-刚刚打破了-ai-的经济学规则/)
- [前沿效率` -> cutting-](https://docs.4allapi.com/en/news/2026-07-29-gpt-5-6-如何融合前沿智能与前沿效率/): GPT-5.6 boosts efficiency in models, inference, and agent
- [Google Flow Music Introdu](https://docs.4allapi.com/en/news/2026-07-29-google-flow-music-中推出-lyria-3-5-在音乐性-歌词-演唱和创作控制方面全面升级/): Lyria
- [5 ways AI mode in Search helps you](https://docs.4allapi.com/en/news/2026-07-28-search-中-ai-模式帮助你享受现实世界的-5-种方式/): Google''s AI search tools help you make the most of offline time,
- [# OlmoEarth 平台：面向行星尺度的地理空间推理](https://docs.4allapi.com/en/news/2026-07-28-olmoearth-平台-面向行星尺度的地理空间推理/): Ai2''s latest blog on Hugging Face focuses on sharing
- [LFM2.5 Encoder](https://docs.4allapi.com/en/news/2026-07-28-lfm2-5-编码器-在-cpu-上实现快速长上下文推理/): Liquid AI''s blog compares LFMs to Transformers,
- [Gemini Robotics 2 brings whole](https://docs.4allapi.com/en/news/2026-07-28-gemini-robotics-2-为机器人带来全身智能/): Robots are being trained for whole-body control, dexterous
- [Gemini API Hosted Agents:](https://docs.4allapi.com/en/news/2026-07-28-gemini-api-托管智能体-3-6-flash-hooks-等/): Gemini API''s Managed Agents get new capabilities to
- [Codex from 0 to](https://docs.4allapi.com/en/news/2026-07-28-codex-from-0-to-10m-users-building-chatgpt-work-akshay-nathan-openai/): OpenAI''s lead engineer shares how ChatGPT Work uses subagents and memory''s core product engineering lead on how they are building ChatGPT Work to make AGI accessible to all of humanity: Sites, OpenClaw, Memory, Subagents, Finance, No-Code and advice.
- [5 Ways to Host the Ultimate](https://docs.4allapi.com/en/news/2026-07-28-5种用-google-搜索举办终极晚宴派对的方法/): AI cuts costs and boosts efficiency in party planning by generating menus, designing tablescap
- [Anatomy](https://docs.4allapi.com/en/news/2026-07-27-前沿实验室智能体入侵剖析-2026年7月事件的技术时间线/): Open source and open science democratize AI by lowering innovation barriers and fostering global
- [into)](https://docs.4allapi.com/en/news/2026-07-27-nvidia-cosmos-h-dreams-将实时生成式仿真带入外科机器人领域/): NVIDIA''s Hugging Face blog highlights its HW/SW stack
- [[AINews](https://docs.4allapi.com/en/news/2026-07-25-ainews-claude-opus-5-以-opus-价格实现-fable-级性能-仅为-fable-的一半/): Anthropic leads the Fable distillation task, showcasing its edge in
- [、](https://docs.4allapi.com/en/news/2026-07-24-ainews-black-forest-labs-flux-3-超越-seedance-2-0-gemini-omni-和-grok-imagine-的多模态流/): BFL''s major win validates its AI chip competitiveness, boosting future funding
- [# Bringing Nunchaku 4](https://docs.4allapi.com/en/news/2026-07-23-将-nunchaku-4-位扩散推理引入-diffusers/): Open source and open science democratize AI by sharing models and data, lowering