# UltraGPT API > OpenAI- and Anthropic-compatible API for 126 models (chat, embeddings, images, speech, transcription, video, realtime, rerank, search). One API key, one prepaid balance, billed per token at the provider's price. Base URL: https://api.ultragpt.pro/v1 Auth: `Authorization: Bearer ugpt_live_...` (or `x-api-key` for Anthropic SDKs). Use the official OpenAI SDK with `base_url` set to the base URL, or the Anthropic SDK against /v1/messages. ## Docs - [Quickstart](https://developers.ultragpt.pro/docs/quickstart): Make your first UltraGPT API request in under a minute: create a key, set the base URL in the OpenAI SDK (Python, Node.js) or curl, and call any model. - [Authentication](https://developers.ultragpt.pro/docs/authentication): Authenticate to the UltraGPT API with a Bearer key or x-api-key, and lock keys down with IP and domain allowlists, model allowlists, rotation and expiry. - [SDKs & tools](https://developers.ultragpt.pro/docs/sdks): Use UltraGPT with the OpenAI SDK, LangChain, LlamaIndex, the Vercel AI SDK, Cursor, Cline, Continue, Codex CLI, Open WebUI, n8n and other OpenAI-compatible tools. - [Anthropic SDK & Claude Code](https://developers.ultragpt.pro/docs/anthropic): Call Claude and other models through the Anthropic Messages API (/v1/messages) with the official Anthropic SDK, and run Claude Code on UltraGPT with ANTHROPIC_BASE_URL. - [Migrate to UltraGPT](https://developers.ultragpt.pro/docs/migrate): Move from OpenAI, Anthropic or OpenRouter to UltraGPT by changing the base URL and key. Parameter, model-name and header mapping included. - [Cookbook](https://developers.ultragpt.pro/docs/cookbook): Copy-paste recipes for the UltraGPT API: chatbots, RAG, structured extraction, agents with tools, image generation and more. - [Chat completions](https://developers.ultragpt.pro/docs/chat): POST /v1/chat/completions on UltraGPT: messages, every supported parameter (temperature, max_tokens, top_p, stop, seed…) and the response format, OpenAI-compatible. - [Responses API](https://developers.ultragpt.pro/docs/responses): The OpenAI Responses API (/v1/responses) on UltraGPT: stateful conversations with previous_response_id, tools and streaming, for any model. - [Streaming](https://developers.ultragpt.pro/docs/streaming): Stream tokens from any model with server-sent events (stream: true), get usage and cost in the last chunk, and handle cancellations. - [Tools & JSON](https://developers.ultragpt.pro/docs/tools): Function calling, tool_choice, parallel tool calls, JSON mode and structured outputs with json_schema, across GPT, Claude, Gemini and other models. - [Reasoning](https://developers.ultragpt.pro/docs/reasoning): Control reasoning (thinking) models with reasoning_effort and token budgets, and read reasoning output and reasoning token usage. - [Images & files in](https://developers.ultragpt.pro/docs/vision): Send images and PDF files to vision models through the UltraGPT API, as URLs or base64, with the OpenAI message format. - [Web search](https://developers.ultragpt.pro/docs/web): Give any model live web search with the web plugin or :online, and get answers with URL citations. - [Provider routing](https://developers.ultragpt.pro/docs/routing): Route requests between a model’s endpoints by price, throughput or latency, with order, only, ignore, max_price and automatic failover. - [Prompt caching](https://developers.ultragpt.pro/docs/caching): Cut input costs with prompt caching: automatic caching and cache_control breakpoints for Claude, and cached-input pricing per model. - [Model fallbacks](https://developers.ultragpt.pro/docs/fallbacks): Keep requests working when a model fails: pass a models array and UltraGPT retries the next model automatically, billed only for the one that answers. - [Presets](https://developers.ultragpt.pro/docs/presets): Save model, system prompt and parameters as a versioned preset and call it with model: "@preset/slug" from any client. - [Response caching](https://developers.ultragpt.pro/docs/response-cache): Return identical requests from cache for free with UltraGPT response caching: TTLs, cache keys and cache-status headers. - [Plugins & variants](https://developers.ultragpt.pro/docs/plugins): Response-healing for valid JSON, context-compression for long prompts, :nitro / :floor / :thinking model variants and zero-completion insurance. - [Embeddings](https://developers.ultragpt.pro/docs/embeddings): Create text embeddings for search, RAG and clustering with /v1/embeddings: models, dimensions, batching and pricing. - [Image generation](https://developers.ultragpt.pro/docs/images): Generate and edit images through one API (/v1/images/generations, /v1/images/edits): sizes, aspect ratios, quality and reference images across image models. - [Speech & transcription](https://developers.ultragpt.pro/docs/audio): Text-to-speech with /v1/audio/speech and speech-to-text with /v1/audio/transcriptions: voices, formats, languages and pricing. - [Batch API](https://developers.ultragpt.pro/docs/batch): Run large jobs asynchronously with the Files and Batch API: upload JSONL, track the batch and download results, at a lower price and without rate limits. - [Rerank, moderation & search](https://developers.ultragpt.pro/docs/retrieval): Rerank documents for RAG, moderate content and run web search through dedicated UltraGPT endpoints (/v1/rerank, /v1/moderations, /v1/search). - [Video generation](https://developers.ultragpt.pro/docs/videos): Generate video from text or images with /v1/videos: Sora, Veo, Kling, Seedance, Wan and more, with durations, resolutions and per-second pricing. - [Realtime voice](https://developers.ultragpt.pro/docs/realtime): Build low-latency voice agents over WebSocket with /v1/realtime: sessions, audio formats, tools and audio-token pricing. - [Idempotent retries](https://developers.ultragpt.pro/docs/idempotency): Retry safely with the Idempotency-Key header: the same request is never run or billed twice. - [App attribution](https://developers.ultragpt.pro/docs/attribution): Attribute requests to your app with the X-Title and HTTP-Referer headers and see usage per app. - [Key, credits & generations](https://developers.ultragpt.pro/docs/account): Read your key’s limits, your credit balance and the exact cost of any generation through the API (/v1/key, /v1/credits, /v1/generation). - [Management keys & activity](https://developers.ultragpt.pro/docs/management): Provision and limit API keys programmatically with management keys, and read per-request activity for your customers. - [Webhooks](https://developers.ultragpt.pro/docs/webhooks): Receive signed webhooks for credits, keys, videos and model deprecations, verify the HMAC signature and handle retries. - [Prompt logs, audit & security](https://developers.ultragpt.pro/docs/observability): Opt-in encrypted prompt logging with retention, an audit log of account changes, and email step-up for sensitive actions. - [Bring your own key](https://developers.ultragpt.pro/docs/byok): Bring your own OpenRouter or OpenAI key (BYOK): requests for that provider run on your account and UltraGPT bills only a small fee. - [Organizations](https://developers.ultragpt.pro/docs/teams): Share one balance across a team: organizations, member roles, invites and per-member spend limits. - [Billing](https://developers.ultragpt.pro/docs/billing): How UltraGPT API billing works: prepaid credits, crypto and card top-ups, per-token pricing, usage.cost in every response, and no subscription. - [Rate limits](https://developers.ultragpt.pro/docs/limits): UltraGPT API rate limits: requests and tokens per minute, concurrency, per-key spend limits and how to handle 429 errors. - [Errors](https://developers.ultragpt.pro/docs/errors): Every UltraGPT API error code (400, 401, 402, 403, 404, 410, 429, 502, 503) with its type, cause and whether to retry. - [Model deprecations](https://developers.ultragpt.pro/docs/deprecations): How models are retired on UltraGPT: Deprecation and Sunset headers, automatic replacements and 410 model_retired. - [Changelog](https://developers.ultragpt.pro/docs/changelog): What’s new in the UltraGPT API: new endpoints, models, parameters and behaviour changes, by date. - [API reference](https://developers.ultragpt.pro/docs/reference): every endpoint with parameters, request bodies and responses. - [OpenAPI spec](https://developers.ultragpt.pro/openapi.json) - [Postman collection](https://developers.ultragpt.pro/ultragpt-api.postman_collection.json) ## Model labs (all models and prices per lab) - [Claude API](https://developers.ultragpt.pro/providers/anthropic): Claude Opus, Sonnet and Haiku models from Anthropic. - [OpenAI API (GPT)](https://developers.ultragpt.pro/providers/openai): GPT, o-series, Codex, image, speech and embedding models models from OpenAI. - [Gemini API](https://developers.ultragpt.pro/providers/google): Gemini, Gemma, Imagen and Veo models from Google. - [Grok API](https://developers.ultragpt.pro/providers/xai): Grok models from xAI. - [DeepSeek API](https://developers.ultragpt.pro/providers/deepseek): DeepSeek V and R models from DeepSeek. - [Qwen API](https://developers.ultragpt.pro/providers/alibaba): Qwen and Wan models from Alibaba. - [Kimi API](https://developers.ultragpt.pro/providers/moonshot): Kimi models from Moonshot AI. - [GLM API](https://developers.ultragpt.pro/providers/zhipu): GLM models from Z.ai. - [MiMo API](https://developers.ultragpt.pro/providers/xiaomi): MiMo models from Xiaomi. - [MiniMax API](https://developers.ultragpt.pro/providers/minimax): MiniMax models from MiniMax. - [Seed API (ByteDance)](https://developers.ultragpt.pro/providers/bytedance): Seed and Doubao models from ByteDance. ## Integrations (setup guides) - [Claude Code](https://developers.ultragpt.pro/integrations/claude-code): Anthropic's terminal coding agent on UltraGPT's Claude models, through our Anthropic-compatible /v1/messages. - [Cursor](https://developers.ultragpt.pro/integrations/cursor): Use UltraGPT models in Cursor chat through the OpenAI base-URL override. - [Cline](https://developers.ultragpt.pro/integrations/cline): The open-source coding agent for VS Code, with any UltraGPT model. - [Roo Code](https://developers.ultragpt.pro/integrations/roo-code): Multi-mode coding agent for VS Code (Code, Architect, Ask, Debug). - [Kilo Code](https://developers.ultragpt.pro/integrations/kilo-code): Open-source coding agent for VS Code and JetBrains. - [OpenAI Codex CLI](https://developers.ultragpt.pro/integrations/codex-cli): OpenAI's terminal coding agent, talking to our Responses API. - [Continue](https://developers.ultragpt.pro/integrations/continue): Open-source AI assistant for VS Code and JetBrains: chat, edit, autocomplete. - [Aider](https://developers.ultragpt.pro/integrations/aider): AI pair programming in your terminal, committing straight to git. - [opencode](https://developers.ultragpt.pro/integrations/opencode): Open-source terminal coding agent with a custom-provider config. - [Zed](https://developers.ultragpt.pro/integrations/zed): Zed's Agent Panel and inline assist with UltraGPT models. - [Goose](https://developers.ultragpt.pro/integrations/goose): Block's open-source on-machine agent (CLI and desktop). - [Claude Desktop](https://developers.ultragpt.pro/integrations/claude-desktop): The Claude app for macOS and Windows (chat, Cowork, Code) on UltraGPT's Claude models, as a third-party inference gateway. - [Open WebUI](https://developers.ultragpt.pro/integrations/open-webui): Self-hosted ChatGPT-style interface; all UltraGPT models show up in its model list. - [LibreChat](https://developers.ultragpt.pro/integrations/librechat): Open-source multi-provider chat, configured as a custom endpoint. - [LobeChat](https://developers.ultragpt.pro/integrations/lobechat): Modern open-source chat UI with plugins and agents. - [Chatbox](https://developers.ultragpt.pro/integrations/chatbox): Desktop and mobile chat client for Windows, macOS, Linux, iOS and Android. - [Jan](https://developers.ultragpt.pro/integrations/jan): Open-source desktop assistant that mixes local and remote models. - [SillyTavern](https://developers.ultragpt.pro/integrations/sillytavern): Character and roleplay front end, through its Custom chat-completion source. - [TypingMind](https://developers.ultragpt.pro/integrations/typingmind): Polished chat front end; add UltraGPT models as custom models. - [Obsidian Copilot](https://developers.ultragpt.pro/integrations/obsidian): Chat with your Obsidian vault using the Copilot plugin. - [OpenAI SDK](https://developers.ultragpt.pro/integrations/openai-sdk): The official OpenAI libraries work unchanged: set base_url and the key. - [Anthropic SDK](https://developers.ultragpt.pro/integrations/anthropic-sdk): Anthropic's Python and TypeScript SDKs against /v1/messages, for every chat model. - [Vercel AI SDK](https://developers.ultragpt.pro/integrations/vercel-ai-sdk): TypeScript toolkit for AI apps (generateText, streamText, tools, useChat). - [LangChain](https://developers.ultragpt.pro/integrations/langchain): Use ChatOpenAI / OpenAIEmbeddings with UltraGPT in LangChain and LangGraph. - [LlamaIndex](https://developers.ultragpt.pro/integrations/llamaindex): RAG and agents over your data, with OpenAILike for the LLM. - [OpenAI Agents SDK](https://developers.ultragpt.pro/integrations/openai-agents-sdk): Multi-agent workflows with handoffs and guardrails, on any UltraGPT model. - [Pydantic AI](https://developers.ultragpt.pro/integrations/pydantic-ai): Type-safe Python agents with structured, validated outputs. - [CrewAI](https://developers.ultragpt.pro/integrations/crewai): Role-playing multi-agent crews; every agent can use a different UltraGPT model. - [LiteLLM](https://developers.ultragpt.pro/integrations/litellm): Call UltraGPT through the LiteLLM SDK, or put it behind a LiteLLM proxy. - [Mastra](https://developers.ultragpt.pro/integrations/mastra): TypeScript agent framework; it takes any AI SDK model. - [DSPy](https://developers.ultragpt.pro/integrations/dspy): Program (not prompt) language models, and optimize the prompts automatically. - [Instructor](https://developers.ultragpt.pro/integrations/instructor): Structured outputs as Pydantic models, with validation and retries. - [smolagents](https://developers.ultragpt.pro/integrations/smolagents): Hugging Face's minimal code-writing agents. - [AutoGen](https://developers.ultragpt.pro/integrations/autogen): Microsoft's multi-agent framework (AgentChat, v0.4+). - [Haystack](https://developers.ultragpt.pro/integrations/haystack): deepset's pipelines for RAG and agents. - [Semantic Kernel](https://developers.ultragpt.pro/integrations/semantic-kernel): Microsoft's SDK for AI agents in C#, Python and Java. - [Spring AI](https://developers.ultragpt.pro/integrations/spring-ai): ChatClient for Spring Boot apps through the OpenAI starter. - [n8n](https://developers.ultragpt.pro/integrations/n8n): Workflow automation: AI Agent, chains and the OpenAI nodes on UltraGPT. - [Dify](https://developers.ultragpt.pro/integrations/dify): LLM app builder: chatflows, agents and RAG knowledge bases. - [Flowise](https://developers.ultragpt.pro/integrations/flowise): Drag-and-drop builder for LLM flows and agents. - [Langflow](https://developers.ultragpt.pro/integrations/langflow): Visual builder for agents and RAG on top of LangChain. - [Make](https://developers.ultragpt.pro/integrations/make): Call UltraGPT from any Make scenario with the HTTP module. ## Models - [grok-4.7](https://developers.ultragpt.pro/models/grok-4.7) (chat, 500000 ctx): $2/M in, $6/M out - [gpt-6-astra](https://developers.ultragpt.pro/models/gpt-6-astra) (chat, 1050000 ctx): $10/M in, $50/M out - [gpt-6-astra-pro](https://developers.ultragpt.pro/models/gpt-6-astra-pro) (chat, 1050000 ctx): $10/M in, $50/M out - [gpt-6.1-sol](https://developers.ultragpt.pro/models/gpt-6.1-sol) (chat, 1050000 ctx): $2/M in, $10/M out - [gpt-6.1-terra](https://developers.ultragpt.pro/models/gpt-6.1-terra) (chat, 1050000 ctx): $1/M in, $5/M out - [gpt-5.6-luna](https://developers.ultragpt.pro/models/gpt-5.6-luna) (chat, 1050000 ctx): $0.2/M in, $1.2/M out - [gpt-5.6-terra](https://developers.ultragpt.pro/models/gpt-5.6-terra) (chat, 1050000 ctx): $2/M in, $12/M out - [gpt-6-luna](https://developers.ultragpt.pro/models/gpt-6-luna) (chat, 1050000 ctx): $0.1/M in, $0.5/M out - [gpt-6-luna-pro](https://developers.ultragpt.pro/models/gpt-6-luna-pro) (chat, 1050000 ctx): $0.1/M in, $0.5/M out - [claude-opus-4.6](https://developers.ultragpt.pro/models/claude-opus-4.6) (chat, 1000000 ctx): $5/M in, $25/M out - [claude-opus-4.8](https://developers.ultragpt.pro/models/claude-opus-4.8) (chat, 1000000 ctx): $5/M in, $25/M out - [gemini-3.8-flash](https://developers.ultragpt.pro/models/gemini-3.8-flash) (chat, 1048576 ctx): $0.75/M in, $3.75/M out - [gemini-4-argon](https://developers.ultragpt.pro/models/gemini-4-argon) (chat, 1048576 ctx): $2/M in, $10/M out - [gemini-4-carbon](https://developers.ultragpt.pro/models/gemini-4-carbon) (chat, 1048576 ctx): $2/M in, $10/M out - [gemini-4-hydrogen](https://developers.ultragpt.pro/models/gemini-4-hydrogen) (chat, 1048576 ctx): $1.5/M in, $7.5/M out - [ai-voice](https://developers.ultragpt.pro/models/ai-voice) (speech): - [gemini-3.7-flash](https://developers.ultragpt.pro/models/gemini-3.7-flash) (chat, 1048576 ctx): $0.75/M in, $3.75/M out - [seedance-2-0-mini](https://developers.ultragpt.pro/models/seedance-2-0-mini) (video): $0.015/second - [gpt-5.6-luna-pro](https://developers.ultragpt.pro/models/gpt-5.6-luna-pro) (chat, 1050000 ctx): $0.2/M in, $1.2/M out - [grok-imagine-image-2.0](https://developers.ultragpt.pro/models/grok-imagine-image-2.0) (image): ~$0.06/image - [command-a-plus](https://developers.ultragpt.pro/models/command-a-plus) (chat, 192000 ctx): $0.3/M in, $1.5/M out - [glm-5.2](https://developers.ultragpt.pro/models/glm-5.2) (chat): $0.144/M in, $7/M out - [glm-5.3](https://developers.ultragpt.pro/models/glm-5.3) (chat, 1048576 ctx): $0.144/M in, $4.8/M out - [glm-5.3-flash](https://developers.ultragpt.pro/models/glm-5.3-flash) (chat, 1048576 ctx): $0.15/M in, $0.5/M out - [glm-5.3-prime](https://developers.ultragpt.pro/models/glm-5.3-prime) (chat, 1000000 ctx): $1.68/M in, $5.28/M out - [hy3](https://developers.ultragpt.pro/models/hy3) (chat, 262144 ctx): $0.132/M in, $0.528/M out - [inkling](https://developers.ultragpt.pro/models/inkling) (chat, 524288 ctx): $1/M in, $4.05/M out - [longcat-2.0](https://developers.ultragpt.pro/models/longcat-2.0) (chat, 1048756 ctx): $0.3/M in, $1.2/M out - [mercury-2.5](https://developers.ultragpt.pro/models/mercury-2.5) (chat, 260000 ctx): $0.04/M in, $0.15/M out - [mimo-v2.6-pro](https://developers.ultragpt.pro/models/mimo-v2.6-pro) (chat, 1048576 ctx): $0.435/M in, $0.87/M out - [muse-spark-1.3](https://developers.ultragpt.pro/models/muse-spark-1.3) (chat, 1048576 ctx): $1.25/M in, $4.25/M out - [nemotron-3-ultra](https://developers.ultragpt.pro/models/nemotron-3-ultra) (chat, 262144 ctx): $0.5/M in, $2.2/M out - [qwen-image-3](https://developers.ultragpt.pro/models/qwen-image-3) (image): ~$0.03/image - [seed-2.0-code](https://developers.ultragpt.pro/models/seed-2.0-code) (chat, 262144 ctx): $0.5/M in, $3/M out - [seed-2.1-turbo](https://developers.ultragpt.pro/models/seed-2.1-turbo) (chat, 262144 ctx): $0.5/M in, $2.5/M out - [mimo-v2.6-flash](https://developers.ultragpt.pro/models/mimo-v2.6-flash) (chat, 1048576 ctx): $0.14/M in, $0.28/M out - [gpt-5.4-mini](https://developers.ultragpt.pro/models/gpt-5.4-mini) (chat, 400000 ctx): $0.75/M in, $4.5/M out - [mai-image-2.6](https://developers.ultragpt.pro/models/mai-image-2.6) (image): ~$0.0707/image - [mai-image-2.6-flash](https://developers.ultragpt.pro/models/mai-image-2.6-flash) (image): ~$0.02/image - [claude-haiku-5.5](https://developers.ultragpt.pro/models/claude-haiku-5.5) (chat, 1000000 ctx): $0.125/M in, $0.5/M out - [claude-opus-5](https://developers.ultragpt.pro/models/claude-opus-5) (chat): $5/M in, $25/M out - [claude-opus-5.5](https://developers.ultragpt.pro/models/claude-opus-5.5) (chat, 1000000 ctx): $4/M in, $20/M out - [claude-sonnet-4.6](https://developers.ultragpt.pro/models/claude-sonnet-4.6) (chat, 1000000 ctx): $3/M in, $15/M out - [claude-sonnet-5](https://developers.ultragpt.pro/models/claude-sonnet-5) (chat, 1000000 ctx): $2/M in, $10/M out - [claude-sonnet-5.5](https://developers.ultragpt.pro/models/claude-sonnet-5.5) (chat, 1000000 ctx): $2/M in, $10/M out - [gpt-oss-120b](https://developers.ultragpt.pro/models/gpt-oss-120b) (chat, 131072 ctx): $0.15/M in, $0.6/M out - [mai-voice-2](https://developers.ultragpt.pro/models/mai-voice-2) (speech): - [muse-image](https://developers.ultragpt.pro/models/muse-image) (image): ~$0.01/image - [gemini-3.1-flash-lite-image](https://developers.ultragpt.pro/models/gemini-3.1-flash-lite-image) (image, 65536 ctx): ~$0.0339/image - [gemini-nano-banana-2.1](https://developers.ultragpt.pro/models/gemini-nano-banana-2.1) (image, 65536 ctx): ~$0.04/image - [seedance-2-0-fast](https://developers.ultragpt.pro/models/seedance-2-0-fast) (video): $0.018/second - [grok-4.6](https://developers.ultragpt.pro/models/grok-4.6) (chat, 500000 ctx): $2/M in, $6/M out - [kimi-k2.7-code](https://developers.ultragpt.pro/models/kimi-k2.7-code) (chat, 262144 ctx): $0.6712/M in, $3.35/M out - [seedance-2-0](https://developers.ultragpt.pro/models/seedance-2-0) (video): $0.033/second - [gemini-3.1-pro](https://developers.ultragpt.pro/models/gemini-3.1-pro) (chat, 1048576 ctx): $2/M in, $12/M out - [gemini-3.1-flash-lite](https://developers.ultragpt.pro/models/gemini-3.1-flash-lite) (chat, 1048576 ctx): $0.25/M in, $1.5/M out - [seedance-2-5](https://developers.ultragpt.pro/models/seedance-2-5) (video): $0.046/second - [grok-4.5](https://developers.ultragpt.pro/models/grok-4.5) (chat, 500000 ctx): $2/M in, $6/M out - [veo-3-1-lite](https://developers.ultragpt.pro/models/veo-3-1-lite) (video): $0.05/second - [deepgram-aura-2](https://developers.ultragpt.pro/models/deepgram-aura-2) (speech): - [wan-3-0](https://developers.ultragpt.pro/models/wan-3-0) (video): $0.05/second - [grok-imagine-video-1-5](https://developers.ultragpt.pro/models/grok-imagine-video-1-5) (video): $0.08/second - [claude-opus-4.7](https://developers.ultragpt.pro/models/claude-opus-4.7) (chat, 1000000 ctx): $5/M in, $25/M out - [grok-voice](https://developers.ultragpt.pro/models/grok-voice) (speech): - [kling-v3-0](https://developers.ultragpt.pro/models/kling-v3-0) (video): $0.126/second - [claude-haiku-4-5](https://developers.ultragpt.pro/models/claude-haiku-4-5) (chat, 200000 ctx): $1/M in, $5/M out - [claude-fable-5](https://developers.ultragpt.pro/models/claude-fable-5) (chat, 1000000 ctx): $10/M in, $50/M out - [claude-fable-5.1](https://developers.ultragpt.pro/models/claude-fable-5.1) (chat, 1000000 ctx): $10/M in, $50/M out - [claude-fable-5.2](https://developers.ultragpt.pro/models/claude-fable-5.2) (chat, 1000000 ctx): $8/M in, $40/M out - [veo-3-1-fast](https://developers.ultragpt.pro/models/veo-3-1-fast) (video): $0.1/second - [deepseek-v4-pro](https://developers.ultragpt.pro/models/deepseek-v4-pro) (chat, 1048576 ctx): $0.9483/M in, $1.8966/M out - [deepseek-v4.1-flash](https://developers.ultragpt.pro/models/deepseek-v4.1-flash) (chat, 1048576 ctx): $0.3/M in, $1.2/M out - [deepseek-v4.1-flash-1017](https://developers.ultragpt.pro/models/deepseek-v4.1-flash-1017) (chat, 1048576 ctx): $0.036/M in, $0.5/M out - [deepseek-v4-flash](https://developers.ultragpt.pro/models/deepseek-v4-flash) (chat, 1048576 ctx): $0.019/M in, $1.28/M out - [minimax-m3.1-flash](https://developers.ultragpt.pro/models/minimax-m3.1-flash) (chat, 1048576 ctx): $0.15/M in, $0.6/M out - [deepseek-v3.2](https://developers.ultragpt.pro/models/deepseek-v3.2) (chat, 163840 ctx): $0.26/M in, $0.38/M out - [mimo-v2.5](https://developers.ultragpt.pro/models/mimo-v2.5) (chat, 1050000 ctx): $0.14/M in, $0.28/M out - [minimax-speech](https://developers.ultragpt.pro/models/minimax-speech) (speech): - [veo-3](https://developers.ultragpt.pro/models/veo-3) (video): $0.4/second - [flux-3-video](https://developers.ultragpt.pro/models/flux-3-video) (video): $0.17/second - [qwen3.7-max](https://developers.ultragpt.pro/models/qwen3.7-max) (chat, 1000000 ctx): $1.475/M in, $4.425/M out - [qwen3.8-max](https://developers.ultragpt.pro/models/qwen3.8-max) (chat, 1000000 ctx): $2/M in, $6/M out - [qwen3.8-max-prime](https://developers.ultragpt.pro/models/qwen3.8-max-prime) (chat, 1000000 ctx): $4/M in, $12/M out - [qwen3.6-flash](https://developers.ultragpt.pro/models/qwen3.6-flash) (chat, 1000000 ctx): $0.1875/M in, $1.125/M out - [qwen3.8-flash](https://developers.ultragpt.pro/models/qwen3.8-flash) (chat, 1000000 ctx): $0.15/M in, $0.47/M out - [grok-imagine-video-1-5-lite](https://developers.ultragpt.pro/models/grok-imagine-video-1-5-lite) (video): $0.02/second - [grok-imagine-video](https://developers.ultragpt.pro/models/grok-imagine-video) (video): $0.05/second - [seedance-1-5-pro](https://developers.ultragpt.pro/models/seedance-1-5-pro) (video): $0.011/second - [heygen-video-1](https://developers.ultragpt.pro/models/heygen-video-1) (video): $0.02/second - [hailuo-3-max](https://developers.ultragpt.pro/models/hailuo-3-max) (video): $0.05/second - [hailuo-3](https://developers.ultragpt.pro/models/hailuo-3) (video): $0.13/second - [sonar-reasoning-pro](https://developers.ultragpt.pro/models/sonar-reasoning-pro) (chat, 128000 ctx): $2/M in, $8/M out - [gpt-image-2.5-flare](https://developers.ultragpt.pro/models/gpt-image-2.5-flare) (image): ~$0.0172/image - [gpt-image-2.6-flare](https://developers.ultragpt.pro/models/gpt-image-2.6-flare) (image): ~$0.0172/image - [hailuo-2-3](https://developers.ultragpt.pro/models/hailuo-2-3) (video): $0.082/second - [gpt-image-2.5-sunburst](https://developers.ultragpt.pro/models/gpt-image-2.5-sunburst) (image): ~$0.0306/image - [gpt-image-2.6-sunburst](https://developers.ultragpt.pro/models/gpt-image-2.6-sunburst) (image): ~$0.0306/image - [sonar-pro](https://developers.ultragpt.pro/models/sonar-pro) (chat, 200000 ctx): $3/M in, $15/M out - [wan-3-0-prime](https://developers.ultragpt.pro/models/wan-3-0-prime) (video): $0.068/second - [wan-2-7](https://developers.ultragpt.pro/models/wan-2-7) (video): $0.1/second - [gpt-image-2](https://developers.ultragpt.pro/models/gpt-image-2) (image): ~$0.018/image - [wan-2-6](https://developers.ultragpt.pro/models/wan-2-6) (video): $0.1/second - [grok-imagine-image-quality](https://developers.ultragpt.pro/models/grok-imagine-image-quality) (image): ~$0.05/image - [happyhorse-1-1](https://developers.ultragpt.pro/models/happyhorse-1-1) (video): $0.099/second - [happyhorse-1-0](https://developers.ultragpt.pro/models/happyhorse-1-0) (video): $0.099/second - [kling-video-o1](https://developers.ultragpt.pro/models/kling-video-o1) (video): $0.112/second - [gemini-3-pro-image-preview](https://developers.ultragpt.pro/models/gemini-3-pro-image-preview) (image, 65536 ctx): ~$0.1374/image - [kling-v3-0-pro](https://developers.ultragpt.pro/models/kling-v3-0-pro) (video): $0.168/second - [runway-gen-4-5](https://developers.ultragpt.pro/models/runway-gen-4-5) (video): $0.12/second - [flux.2-pro](https://developers.ultragpt.pro/models/flux.2-pro) (image): ~$0.03/image - [audio-transcribe](https://developers.ultragpt.pro/models/audio-transcribe) (transcription): - [deepseek-v4-flash-vision-exp](https://developers.ultragpt.pro/models/deepseek-v4-flash-vision-exp) (chat, 1048576 ctx): $0.44/M in, $1.32/M out - [qwen3-embedding-8b](https://developers.ultragpt.pro/models/qwen3-embedding-8b) (embedding, 32768 ctx): $0.01/M in, $0/M out - [text-embedding-3-large](https://developers.ultragpt.pro/models/text-embedding-3-large) (embedding, 8192 ctx): $0.13/M in, $0/M out - [text-embedding-3-small](https://developers.ultragpt.pro/models/text-embedding-3-small) (embedding, 8192 ctx): $0.02/M in, $0/M out - [cohere-rerank-v3.5](https://developers.ultragpt.pro/models/cohere-rerank-v3.5) (rerank): $0.002/request - [cohere-rerank-4-fast](https://developers.ultragpt.pro/models/cohere-rerank-4-fast) (rerank): $0.002/request - [cohere-rerank-4-pro](https://developers.ultragpt.pro/models/cohere-rerank-4-pro) (rerank): $0.0025/request - [dataforseo-search](https://developers.ultragpt.pro/models/dataforseo-search) (search): $0.003/request - [exa_ai-search](https://developers.ultragpt.pro/models/exa_ai-search) (search): $0.025/request - [omni-moderation-latest](https://developers.ultragpt.pro/models/omni-moderation-latest) (moderation): $0/request - [perplexity-search](https://developers.ultragpt.pro/models/perplexity-search) (search): $0.005/request - [qwen3-reranker-8b](https://developers.ultragpt.pro/models/qwen3-reranker-8b) (rerank): $0.0005/request - [tavily-search](https://developers.ultragpt.pro/models/tavily-search) (search): $0.008/request - [tavily-search-advanced](https://developers.ultragpt.pro/models/tavily-search-advanced) (search): $0.016/request - [voyage-rerank-2.5](https://developers.ultragpt.pro/models/voyage-rerank-2.5) (rerank): $0.001/request