# MixRoute Complete Documentation > MixRoute is an AI API gateway. One OpenAI-compatible API for 197 models from OpenAI, Anthropic, Google, DeepSeek, xAI, Alibaba, Moonshot, MiniMax, Zhipu, and others. Zero markup pass-through pricing (no platform fee on top of provider rates). Deposit-tier bonuses up to 8% on self-serve. Enterprise tier ($30K+ single top-up) unlocks Reserve Capacity for high-demand image models, custom routing, and dedicated support. Self-serve payment via Visa, Mastercard, JCB, USDT, and USDC. Enterprise via bank wire with invoicing. OpenAI SDK compatible. Migration is a one-line base URL change to `https://api.mixroute.ai/v1`. Version: 1.1.1 Last Updated: May 27, 2026 For the curated short index, see [llms.txt](https://mixroute.ai/llms.txt). For interactive docs, see [docs.mixroute.ai](https://docs.mixroute.ai). For the live model catalog with current rates, see [console.mixroute.ai/models](https://console.mixroute.ai/models). --- # Introduction MixRoute is a professional AI API platform built on the OpenAI API standard. With one API key, developers and teams get access to 197 models across every major provider through a single OpenAI-compatible endpoint. ## How MixRoute Works The base URL is `https://api.mixroute.ai/v1`. Authentication is a Bearer token (`Authorization: Bearer sk-...`). Every request goes to the same endpoint regardless of which model is being called. The `model` field in the request body selects which provider model handles the request. MixRoute translates between the OpenAI request format and the upstream provider's native format where they differ (Anthropic, Gemini, etc.). ## The Business Model MixRoute is pass-through pricing. The platform charges no markup on token costs. Every dollar deposited converts to tokens at the official provider rates. The business runs on cloud reseller margins through upstream providers and on deposit-tier bonuses, not on a tax against token spend. Self-serve customers operate on the same provider rate limits as direct integrations. The advantages are: unified API surface, unified billing, simplified integration, deposit bonuses, and access to providers (like Chinese model labs) that may be difficult to onboard with directly. Enterprise customers ($30,000+ single top-up) gain access to Reserve Capacity, a separate enterprise product where MixRoute pre-commits large-scale reserved throughput with upstream providers and pools it across the enterprise customer base. ## Core Properties - **One Account**: Manage all AI services through a single console. - **One API Standard**: Fully compatible with the OpenAI API format. - **One API Key**: Access every model from every supported provider. - **Zero Markup**: Token-based billing at official provider prices. - **Deposit-Tier Bonuses**: Up to 8% on self-serve, custom bonus on enterprise. - **Reserve Capacity (Enterprise)**: Pre-committed pool with upstream providers for select high-demand image models. - **Crypto-Native**: USDT and USDC deposits with no KYC. - **8 Localized Languages**: English, Simplified Chinese, Traditional Chinese, French, German, Japanese, Russian, Vietnamese. ## Base URL ``` https://api.mixroute.ai/v1 ``` --- # Pricing MixRoute uses a prepaid pass-through model. Credit is loaded into the account and each API request deducts based on actual token usage at official provider prices. No platform fee on top of provider rates. ## Deposit-Tier Bonus Structure | Tier | Single top-up range | Bonus | Maximum bonus | Included features | | ----------- | -------------------- | -------- | ------------- | -------------------------------------------------------------------------------- | | Starter | $10 to $999 | None | None | Access to all 197 models. OpenAI-compatible API. Standard support. | | Growth | $1,000 to $4,999 | 3% | $150 | Everything in Starter, plus priority request queue and advanced usage analytics. | | Pro | $5,000 to $14,999 | 5% | $750 | Higher rate limits, multi-model routing, team management. | | Scale | $15,000 to $29,999 | 8% | $2,400 | Highest rate limits, early access to new models, custom routing strategies. | | Enterprise | $30,000+ single top-up | Custom | Up to 25% | Full enterprise package: Reserve Capacity eligibility, custom routing, dedicated manager, 24/7 support, service contract. | ## Current Promotion **First top-up: 100% bonus.** New users get a 100% bonus on their first top-up between $30 and $500 USD. Maximum bonus: $500. Offer ends May 31, 2026. The first-deposit bonus stacks with the regular tier bonus. Example: a first-time $1,000 deposit earns the $1,000 first-deposit bonus plus the Growth tier 3% bonus on top. ## Payment Methods | Tier | Methods | | ------------ | ---------------------------------------------------------------------------------------------------------------------- | | Self-serve | Credit cards (Visa, Mastercard, JCB) and cryptocurrency (USDT, USDC). No KYC for crypto. | | Enterprise | Bank wire transfer with invoice. Flexible payment terms. Service contract included. | ## Balance Policy Credits do not expire. Whether you top up through self-serve or an enterprise plan, your balance remains valid indefinitely. Use the API at your own pace. ## Per-Model Pricing Per-model token rates are pass-through at official provider pricing. View the live model catalog with current rates at [console.mixroute.ai/models](https://console.mixroute.ai/models). --- # Reserve Capacity (Enterprise Feature) Reserve Capacity is an enterprise-tier product available to customers with single top-ups of $30,000 or more. It is currently offered for high-demand Google image models. ## What Reserve Capacity Is Most AI APIs run on a shared rate-limit pool. When traffic spikes, your requests wait in line. Reserve Capacity is different: MixRoute pre-commits large-scale reserved throughput with upstream providers and pools it across the enterprise customer base. Enterprise customers access dedicated TPM (tokens per minute) at preferential rates without needing to negotiate multi-year commits with the upstream provider directly. ## What Reserve Capacity Provides - **High TPM ceiling**: Production-grade contracted TPM floor sized for sustained enterprise workloads, not best-effort on-demand. - **Dedicated capacity pool**: Isolated from public traffic. Requests run on capacity reserved for enterprise tier, separated from shared rate-limit pools. - **Global capacity pool**: Aggregated demand across regions lets MixRoute commit to volumes individual teams cannot reach alone. Enterprise customers access enterprise-tier pricing without the enterprise-tier commitment. - **Tier-1 enterprise compute**: Same node class as direct contract customers, isolated from public pools. - **Sized to your workload**: TPM and RPM allocated to your forecast, adjustable quarterly. - **Zero procurement overhead**: No multi-year commit, no quota application, no upstream contract management. ## Models Currently Under Reserve Capacity - Google "Nano Banana Pro" (API ID: `gemini-3-pro-image-preview`) - Google "Nano Banana 2" (API ID: `gemini-3.1-flash-image-preview`) ## Capacity Windows by Region Reserve Capacity operates in scheduled windows by region. Peak windows have the highest contracted TPM floors. ### Nano Banana 2 (`gemini-3.1-flash-image-preview`) | Region | Available window | Peak TPM | | ----------- | -------------------------- | --------- | | Los Angeles | 04:00 to 18:00 PDT | 66M TPM | | New York | 07:00 to 21:00 EDT | 66M TPM | | London | 12:00 to 02:00+1 BST | 66M TPM | | Paris | 13:00 to 03:00+1 CEST | 66M TPM | | Singapore | 19:00 to 09:00+1 SGT | 66M TPM | Peak windows shift outside the open/close hours. Ramp-up periods at the start and end of each window typically operate at 48M to 60M TPM. ### Nano Banana Pro (`gemini-3-pro-image-preview`) | Region | Available window | Peak TPM | | ----------- | -------------------------- | ------------- | | Los Angeles | 09:00 to 18:00 PDT | 15-18M TPM | | New York | 12:00 to 21:00 EDT | 15-18M TPM | | London | 17:00 to 02:00+1 BST | 15-18M TPM | | Paris | 18:00 to 03:00+1 CEST | 15-18M TPM | | Singapore | 00:00+1 to 09:00+1 SGT | 15-18M TPM | All times use daylight saving (DST). During standard time, New York, London, and Paris each shift back by 1 hour. The "+1" indicates the time has crossed into the next day in the local time zone. ## Access Reserve Capacity is unlocked via Talk to Sales. Single top-ups of $30,000 or more qualify for enterprise onboarding. The MixRoute team typically reaches out within 24 hours with a tailored quote, contract, and capacity allocation. [Talk to Sales](https://mixroute.ai/talk-to-sales/) --- # Quick Start Get up and running in three steps. ## Step 1: Create Your Account Visit [api.mixroute.ai](https://api.mixroute.ai) (or [console.mixroute.ai](https://console.mixroute.ai)) and click **Sign Up**. 1. Enter your email address, set a password, and complete email verification. 2. Go to **Wallet Management** and complete your initial top-up. Self-serve supports Visa, Mastercard, JCB, USDT, and USDC. ## Step 2: Create an API Key 1. Click **Token Management** in the console navigation. 2. Click **Add Token**. 3. Click **Submit** to generate a key. Your key begins with `sk-`. ## Step 3: Make Your First Call ### Connection Configuration | Configuration | Value | | ------------------- | -------------------------------- | | API URL (Base URL) | `https://api.mixroute.ai/v1` | | API Key | Your created token (e.g. `sk-...`) | | Authentication | Bearer token | | Request Format | Fully OpenAI API compatible | ### Test Call (curl) ```bash curl https://api.mixroute.ai/v1/chat/completions \ -H "Content-Type: application/json" \ -H "Authorization: Bearer YOUR_API_KEY" \ -d '{ "model": "claude-opus-4-7", "messages": [{"role": "user", "content": "Hello!"}] }' ``` ### Python ```python from openai import OpenAI client = OpenAI( api_key="YOUR_API_KEY", base_url="https://api.mixroute.ai/v1" ) response = client.chat.completions.create( model="claude-opus-4-7", messages=[{"role": "user", "content": "Hello!"}] ) print(response.choices[0].message.content) ``` ### Node.js / TypeScript ```javascript import OpenAI from 'openai'; const client = new OpenAI({ apiKey: 'YOUR_API_KEY', baseURL: 'https://api.mixroute.ai/v1' }); const response = await client.chat.completions.create({ model: 'claude-opus-4-7', messages: [{ role: 'user', content: 'Hello!' }] }); console.log(response.choices[0].message.content); ``` ### Go ```go package main import ( "context" "fmt" openai "github.com/sashabaranov/go-openai" ) func main() { config := openai.DefaultConfig("YOUR_API_KEY") config.BaseURL = "https://api.mixroute.ai/v1" client := openai.NewClientWithConfig(config) resp, _ := client.CreateChatCompletion( context.Background(), openai.ChatCompletionRequest{ Model: "claude-opus-4-7", Messages: []openai.ChatCompletionMessage{ {Role: "user", Content: "Hello!"}, }, }, ) fmt.Println(resp.Choices[0].Message.Content) } ``` ### Java ```java OpenAiService service = new OpenAiService( "YOUR_API_KEY", Duration.ofSeconds(60), "https://api.mixroute.ai/v1" ); ChatCompletionRequest request = ChatCompletionRequest.builder() .model("claude-opus-4-7") .messages(List.of( new ChatMessage(ChatMessageRole.USER, "Hello!") )) .build(); ChatCompletionResult result = service.createChatCompletion(request); System.out.println(result.getChoices().get(0).getMessage().getContent()); ``` --- # Complete Model Catalog This catalog reflects the live `/v1/models` API as of May 27, 2026. 197 models total. Use the canonical Model ID exactly as listed when making API requests. For the live model catalog with current per-model rates, see [console.mixroute.ai/models](https://console.mixroute.ai/models). ## OpenAI OpenAI models accessed via the OpenAI API format. ### GPT-5 Series | Model ID | Notes | | ------------------------------ | ------------------------------------------------------------------------ | | `gpt-5.5` | Latest in the GPT-5 lineage. | | `gpt-5.4` | GPT-5.4 base. | | `gpt-5.4-2026-03-05` | GPT-5.4 dated snapshot. | | `gpt-5.4-pro` | GPT-5.4 Pro variant. | | `gpt-5.4-pro-2026-03-05` | Dated snapshot. | | `gpt-5.4-mini` | GPT-5.4 mini. | | `gpt-5.4-mini-2026-03-17` | Dated snapshot. | | `gpt-5.4-nano` | GPT-5.4 nano. | | `gpt-5.3-codex` | GPT-5.3 Codex variant for code generation. | | `gpt-5.3-chat` | GPT-5.3 chat variant. | | `gpt-5.2` | GPT-5.2 base. | | `gpt-5.2-2025-12-11` | Dated snapshot. | | `gpt-5.2-pro` | Pro variant. | | `gpt-5.2-pro-2025-12-11` | Dated snapshot. | | `gpt-5.2-chat` | Chat variant. | | `gpt-5.2-chat-latest` | Pinned-to-latest chat. | | `gpt-5.2-codex` | Codex variant. | | `gpt-5.1` | GPT-5.1 base. | | `gpt-5.1-2025-11-13` | Dated snapshot. | | `gpt-5.1-chat` | Chat variant. | | `gpt-5.1-chat-latest` | Pinned-to-latest. | | `gpt-5.1-codex` | Codex variant. | | `gpt-5.1-codex-max` | Codex max variant. | | `gpt-5.1-codex-mini` | Codex mini variant. | | `gpt-5` | GPT-5 base. | | `gpt-5-2025-08-07` | Dated snapshot. | | `gpt-5-chat` | Chat variant. | | `gpt-5-chat-latest` | Pinned-to-latest. | | `gpt-5-codex` | Codex variant. | | `gpt-5-mini` | GPT-5 mini. | | `gpt-5-mini-2025-08-07` | Dated snapshot. | | `gpt-5-nano` | GPT-5 nano. | | `gpt-5-nano-2025-08-07` | Dated snapshot. | | `gpt-5-pro` | GPT-5 Pro. | | `gpt-5-pro-2025-10-06` | Dated snapshot. | ### GPT-4 Series | Model ID | Notes | | ------------------------------ | ------------------------------------------------------------------------ | | `gpt-4o` | Multimodal model. | | `gpt-4o-2024-11-20` | Dated snapshot. | | `gpt-4o-2024-08-06` | Dated snapshot. | | `gpt-4o-2024-05-13` | Dated snapshot. | | `gpt-4o-mini` | Cost-efficient multimodal. | | `gpt-4o-mini-2024-07-18` | Dated snapshot. | | `gpt-4o-mini-tts` | Text-to-speech variant. | | `gpt-4.1` | GPT-4.1. | | `gpt-4.1-2025-04-14` | Dated snapshot. | | `gpt-4.1-mini` | GPT-4.1 mini. | | `gpt-4.1-mini-2025-04-14` | Dated snapshot. | | `gpt-4.1-nano` | GPT-4.1 nano. | | `gpt-4.1-nano-2025-04-14` | Dated snapshot. | | `gpt-4-turbo` | GPT-4 Turbo. | | `gpt-4-turbo-2024-04-09` | Dated snapshot. | | `gpt-4` | GPT-4 base. | | `gpt-4-0613` | Dated snapshot. | | `gpt-4-0125-preview` | Preview snapshot. | | `gpt-4-1106-preview` | Preview snapshot. | ### GPT-3.5 Series | Model ID | Notes | | ------------------------------ | ------------------------------------------------------------------------ | | `gpt-3.5-turbo` | GPT-3.5 Turbo. | | `gpt-3.5-turbo-0125` | Dated snapshot. | | `gpt-3.5-turbo-1106` | Dated snapshot. | | `gpt-3.5-turbo-16k` | 16K context variant. | ### Reasoning Models (o-Series) | Model ID | Notes | | --------------------------------- | ------------------------------------------------------------------------ | | `o4-mini` | o4-mini reasoning. | | `o4-mini-2025-04-16` | Dated snapshot. | | `o3` | o3 reasoning. | | `o3-2025-04-16` | Dated snapshot. | | `o3-pro` | o3 Pro variant (uses OpenAI Responses endpoint format). | | `o3-pro-2025-06-10` | Dated snapshot. | | `o3-deep-research-2025-06-26` | Deep research variant (uses OpenAI Responses endpoint format). | | `o3-mini` | o3-mini reasoning. | | `o3-mini-2025-01-31` | Dated snapshot. | | `o3-mini-low`, `o3-mini-medium`, `o3-mini-high` | Effort-level variants of o3-mini. | | `o1` | o1 reasoning. | | `o1-2024-12-17` | Dated snapshot. | | `o1-mini` | o1-mini. | | `o1-mini-2024-09-12` | Dated snapshot. | ### Image Generation | Model ID | Notes | | --------------------------------- | ------------------------------------------------------------------------ | | `gpt-image-2` | OpenAI's latest image generation model. | | `gpt-image-1.5` | GPT image 1.5. | | `gpt-image-1` | GPT image 1. | ### Audio | Model ID | Notes | | --------------------------------- | ------------------------------------------------------------------------ | | `whisper-1` | OpenAI Whisper speech-to-text. | ## Anthropic Claude Claude models support both the Anthropic native message format (`/v1/messages`) and the OpenAI Chat Completions format. Thinking variants surface a `reasoning_content` field with the extended reasoning trace. ### Claude 4 Latest Generation | Model ID | Notes | | ---------------------------------------- | ------------------------------------------------------------------ | | `claude-opus-4-7` | Flagship Claude Opus. | | `claude-opus-4-6` | Previous-gen Opus. | | `claude-opus-4-6-thinking` | Opus 4.6 with extended thinking. | | `claude-sonnet-4-6` | Sonnet 4.6, production workhorse. | | `claude-sonnet-4-6-thinking` | Sonnet 4.6 with extended thinking. | | `claude-haiku-4-5` | Fastest Claude tier. | | `claude-haiku-4-5-20251001` | Dated snapshot. | | `claude-haiku-4-5-20251001-thinking` | Haiku with extended thinking. | ### Claude 4.5 Generation | Model ID | Notes | | ---------------------------------------- | ------------------------------------------------------------------ | | `claude-opus-4-5` | Opus 4.5. | | `claude-opus-4-5-20251101` | Dated snapshot. | | `claude-sonnet-4-5` | Sonnet 4.5. | | `claude-sonnet-4-5-20250929` | Dated snapshot. | | `claude-sonnet-4-5-20250929-thinking` | With extended thinking. | ### Claude 4 Earlier | Model ID | Notes | | ---------------------------------------- | ------------------------------------------------------------------ | | `claude-opus-4-1-20250805` | Opus 4.1 dated snapshot. | | `claude-opus-4-20250514` | Opus 4 dated snapshot. | | `claude-sonnet-4-20250514` | Sonnet 4 dated snapshot. | | `claude-sonnet-4-20250514-thinking` | With extended thinking. | ### Claude 3 Series | Model ID | Notes | | ---------------------------------------- | ------------------------------------------------------------------ | | `claude-3-5-sonnet-20240620` | Claude 3.5 Sonnet. | | `claude-3-5-sonnet-20241022` | Claude 3.5 Sonnet refresh. | | `claude-3-7-sonnet-20250219` | Claude 3.7 Sonnet. | ## Google Gemini Gemini models support both Google's native format (`/v1beta/models/{model}:generateContent`) and the OpenAI Chat Completions format. ### Gemini 3 Series (Current) | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `gemini-3.1-pro-preview` | Flagship Gemini 3.1 Pro Preview. | | `gemini-3.1-pro-preview-customtools` | Pro Preview with custom tools. | | `gemini-3.1-flash-image-preview` | Flash Image Preview. Marketed as "Nano Banana 2". | | `gemini-3.1-flash-lite-preview` | Flash Lite Preview. | | `gemini-3.5-flash` | Latest Flash. | | `gemini-3-pro-preview` | Gemini 3 Pro Preview. | | `gemini-3-pro-image-preview` | Pro Image Preview. Marketed as "Nano Banana Pro". | | `gemini-3-flash-preview` | Gemini 3 Flash Preview. | ### Gemini Aliases (Pinned to Latest) | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `gemini-pro-latest` | Always points to the current Gemini Pro. | | `gemini-flash-latest` | Always points to the current Gemini Flash. | | `gemini-flash-lite-latest` | Always points to the current Gemini Flash Lite. | ### Gemini 2.5 Series | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `gemini-2.5-pro` | Gemini 2.5 Pro. | | `gemini-2.5-flash` | Gemini 2.5 Flash. | | `gemini-2.5-flash-image` | Image generation. | | `gemini-2.5-flash-lite` | Flash Lite. | | `gemini-2.5-flash-lite-preview-09-2025` | Preview snapshot. | ### Gemini 2.0 Series | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `gemini-2.0-flash` | Gemini 2.0 Flash. | | `gemini-2.0-flash-001` | Dated snapshot. | | `gemini-2.0-flash-lite` | Flash Lite. | | `gemini-2.0-flash-lite-001` | Dated snapshot. | ### Gemini Specialized | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `gemini-robotics-er-1.5-preview` | Robotics-specialized model. | ### Gemini Embeddings | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `gemini-embedding-001` | Gemini embedding model. | | `gemini-embedding-2-preview` | Gemini embedding v2 preview. | ## DeepSeek | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `deepseek-v4-pro` | DeepSeek's flagship V4 Pro. | | `deepseek-v4-flash` | V4 Flash, lighter and faster variant. | | `deepseek-v3.2` | DeepSeek V3.2. | | `deepseek-v3.2-exp` | V3.2 experimental. | | `deepseek-v3.2-thinking` | V3.2 with extended thinking. | | `deepseek-v3.1` | DeepSeek V3.1. | | `deepseek-v3.1-250821` | Dated snapshot. | | `deepseek-v3.1-terminus` | V3.1 terminus. | | `deepseek-v3` | DeepSeek V3. | | `deepseek-v3-0324` | Dated snapshot. | | `deepseek-v3-2-251201` | Dated snapshot. | | `deepseek-r1` | Reasoning model. Returns `reasoning_content` field with the thinking process. | | `deepseek-r1-0528` | Dated snapshot. | | `deepseek-r1-250528` | Dated snapshot. | ## xAI Grok Grok models support native real-time web search via the `search_parameters` field. | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `grok-4.20-beta-0309-reasoning` | Grok 4.20 with extended reasoning enabled. Recommended default for Grok 4.20. | | `grok-4.20-beta-0309-non-reasoning` | Grok 4.20 without extended reasoning, faster throughput. | | `grok-4.20-multi-agent-beta-0309` | Grok 4.20 multi-agent variant. | | `grok-4-fast-reasoning` | Grok 4 Fast with reasoning. | | `grok-4-fast-non-reasoning` | Grok 4 Fast without reasoning. | | `grok-4-0709` | Previous Grok 4. | | `grok-code-fast-1` | Code-specialized fast model. | | `grok-3` | Grok 3. | | `grok-3-mini` | Grok 3 mini. Supports `reasoning_effort` parameter (low/medium/high). | ## Alibaba Qwen ### Qwen 3.7 Series | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `qwen3.7-max` | Flagship Qwen 3.7 Max. | ### Qwen 3.6 Series | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `qwen3.6-max-preview` | Qwen 3.6 Max Preview. | | `qwen3.6-plus` | Qwen 3.6 Plus. | | `qwen3.6-flash` | Qwen 3.6 Flash. | | `qwen3.6-27b` | 27B parameter variant. | | `qwen3.6-35b-a3b` | 35B activated, 3B active per forward pass (MoE). | ### Qwen 3.5 Series | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `qwen3.5-plus` | Qwen 3.5 Plus. | | `qwen3.5-flash` | Qwen 3.5 Flash. | | `qwen3.5-122b-a10b` | 122B total, 10B active (MoE). | | `qwen3.5-397b-a17b` | 397B total, 17B active (MoE). | | `qwen3.5-35b-a3b` | 35B total, 3B active (MoE). | | `qwen3.5-27b` | 27B parameter variant. | ### Qwen 3 Max | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `qwen3-max` | Qwen 3 Max. | | `qwen3-max-preview` | Qwen 3 Max Preview. | ### Qwen Image Generation | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `qwen-image` | Qwen image generation. | | `qwen-image-2.0` | Qwen image 2.0. | | `qwen-image-2.0-pro` | Qwen image 2.0 Pro. | | `qwen-image-max` | Qwen image Max. | | `qwen-image-plus` | Qwen image Plus. | ### Qwen Image Editing | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `qwen-image-edit` | Image editing. | | `qwen-image-edit-max` | Image editing Max. | | `qwen-image-edit-plus` | Image editing Plus. | ## MiniMax | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `MiniMax-M2` | MiniMax M2. | | `MiniMax-M2.5` | MiniMax M2.5. | | `MiniMax-M2.7` | MiniMax M2.7. | ## Moonshot Kimi ### Kimi K2 Series | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `kimi-k2.6` | Latest Kimi K2. | | `kimi-k2.5` | Previous K2. | | `kimi-k2-thinking` | K2 with extended thinking. | | `kimi-k2-thinking-turbo` | K2 thinking turbo variant. | | `kimi-k2-turbo-preview` | K2 turbo preview. | | `kimi-k2-0711-preview` | Dated K2 preview. | | `kimi-k2-0905-preview` | Dated K2 preview. | ### Moonshot v1 Series | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `moonshot-v1-128k` | 128K context. | | `moonshot-v1-32k` | 32K context. | | `moonshot-v1-8k` | 8K context. | | `moonshot-v1-128k-vision-preview` | 128K with vision. | | `moonshot-v1-32k-vision-preview` | 32K with vision. | | `moonshot-v1-8k-vision-preview` | 8K with vision. | ## Zhipu GLM | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `glm-4.5-air` | Latest GLM. | | `glm-4-plus` | GLM-4 Plus. | | `glm-4-air` | GLM-4 Air. | | `glm-4-airx` | GLM-4 AirX. | | `glm-4-flash` | GLM-4 Flash. | | `glm-4-long` | GLM-4 Long context. | | `glm-4-alltools` | GLM-4 with tool use. | | `glm-4v` | GLM-4 vision. | | `glm-4v-plus` | GLM-4 vision Plus. | | `glm-4` | GLM-4 base. | | `glm-4-0520` | Dated snapshot. | | `glm-3-turbo` | GLM-3 Turbo. | ## Embeddings | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `text-embedding-3-large` | OpenAI's high-quality embedding model. | | `text-embedding-3-small` | OpenAI's cost-efficient embedding. | | `text-embedding-ada-002` | Earlier OpenAI embedding model. | | `gemini-embedding-001` | Google embedding. | | `gemini-embedding-2-preview` | Google embedding v2 preview. | ## Audio | Model ID | Notes | | ---------------------------------------------- | ------------------------------------------------------------------ | | `whisper-1` | OpenAI Whisper transcription. | | `gpt-4o-mini-tts` | Text-to-speech. | --- # API Reference MixRoute exposes multiple API endpoints, all under `https://api.mixroute.ai/v1`. All endpoints use Bearer token authentication. ## Available Endpoints | Endpoint | Path | Description | | ------------------------------------------ | ----------------------------------- | ---------------------------------------------------------------------------- | | Chat Completions (OpenAI) | `/v1/chat/completions` | Universal text chat API for all OpenAI-compatible models. | | Create Responses Request (OpenAI) | `/v1/responses` | Next-generation conversation interface for reasoning models (o3-pro, o3-deep-research). | | Create Messages (Claude) | `/v1/messages` | Claude-native message format. Used by Claude Code and Anthropic clients. | | Count Tokens (Claude) | `/v1/messages/count_tokens` | Calculate token count for Claude messages before sending. | | Gemini Native (Text) | `/v1beta/models/{model}:generateContent` | Use Google Gemini in its native format. | | Gemini Native (Image) | `/v1beta/models/{model}:generateContent` | Gemini image generation in native format. | | Gemini Text Embedding | `/v1beta/models/{model}:embedContent` | Native Gemini embeddings. | | Text Embeddings | `/v1/embeddings` | OpenAI-compatible text-to-vector embeddings. | | Image Generation | `/v1/images/generations` | Text-to-image, image-to-image, image editing. | | Document Reranking | `/v1/rerank` | Rerank documents by relevance for RAG pipelines. | | List Models | `/v1/models` | Get available model information. | ## Chat Completions (OpenAI Format) The primary endpoint. Supports every OpenAI-compatible model in the catalog. ### Authentication Bearer token: `Authorization: Bearer sk-xxxxxxxxxx` ### Request Parameters - `model` (string, required): Canonical model ID from the catalog. - `messages` (array, required): Conversation messages. Each message has `role` (user/system/assistant) and `content`. - `temperature` (number, optional): Randomness control, 0-2. Higher values produce more random responses. - `stream` (boolean, optional): Enable Server-Sent Events streaming. - `max_tokens` (integer, optional): Maximum tokens to generate. - `tools` (array, optional): Function definitions for tool calling. - `tool_choice` (string/object, optional): Control which tool the model uses. - `response_format` (object, optional): JSON Schema for structured output. ### Non-Streaming Example ```bash curl -X POST "https://api.mixroute.ai/v1/chat/completions" \ -H "Content-Type: application/json" \ -H "Authorization: Bearer sk-xxxxxxxxxx" \ -d '{ "model": "claude-opus-4-7", "messages": [ {"role": "system", "content": "You are a helpful assistant"}, {"role": "user", "content": "Briefly introduce artificial intelligence"} ], "temperature": 0.7 }' ``` ### Streaming Example (SSE) ```bash curl -N -X POST "https://api.mixroute.ai/v1/chat/completions" \ -H "Content-Type: application/json" \ -H "Authorization: Bearer sk-xxxxxxxxxx" \ -d '{ "model": "gpt-5.5", "stream": true, "messages": [ {"role": "user", "content": "Tell me a story"} ] }' ``` Streaming chunks end with `data: [DONE]`. ### Tool Calling Example ```bash curl -X POST "https://api.mixroute.ai/v1/chat/completions" \ -H "Content-Type: application/json" \ -H "Authorization: Bearer sk-xxxxxxxxxx" \ -d '{ "model": "gpt-5.5", "messages": [ {"role": "user", "content": "What is the weather in Tokyo?"} ], "tools": [ { "type": "function", "function": { "name": "get_weather", "description": "Get weather information by city", "parameters": { "type": "object", "properties": { "city": {"type": "string"} }, "required": ["city"] } } } ], "tool_choice": "auto" }' ``` ### Structured Output Example ```bash curl -X POST "https://api.mixroute.ai/v1/chat/completions" \ -H "Content-Type: application/json" \ -H "Authorization: Bearer sk-xxxxxxxxxx" \ -d '{ "model": "gpt-5.5", "response_format": { "type": "json_schema", "json_schema": { "name": "Answer", "schema": { "type": "object", "properties": { "summary": {"type": "string"} }, "required": ["summary"] } } }, "messages": [ {"role": "user", "content": "Return a JSON with a summary field"} ] }' ``` ### Reasoning and Thinking Models DeepSeek R1, Qwen QwQ, Gemini Thinking variants, Grok mini, and Claude `-thinking` variants support deep reasoning modes. **DeepSeek R1** returns a `reasoning_content` field with the thinking process: ```bash curl -X POST "https://api.mixroute.ai/v1/chat/completions" \ -H "Content-Type: application/json" \ -H "Authorization: Bearer sk-xxxxxxxxxx" \ -d '{ "model": "deepseek-r1", "messages": [ {"role": "user", "content": "If x^2 + 2x - 3 = 0, find x"} ], "temperature": 0.6 }' ``` **Grok 3 mini** supports `reasoning_effort` parameter (`low` for fast/basic, `medium` for balanced, `high` for deep reasoning). **Claude thinking variants** (e.g., `claude-opus-4-7`, `claude-sonnet-4-6-thinking`, `claude-haiku-4-5-20251001-thinking`) expose Anthropic's extended thinking. Call with the `-thinking` suffix in the model ID. ### Web Search **Claude models** support native web search via the `tools` array: ```json "tools": [ { "type": "web_search_20250305", "name": "web_search", "max_uses": 5 } ] ``` **Grok models** support `search_parameters` with `mode: "auto"` and `return_citations: true`: ```bash curl -X POST "https://api.mixroute.ai/v1/chat/completions" \ -H "Content-Type: application/json" \ -H "Authorization: Bearer sk-xxxxxxxxxx" \ -d '{ "model": "grok-3", "messages": [ {"role": "user", "content": "What are the latest tech news?"} ], "search_parameters": { "mode": "auto", "return_citations": true } }' ``` **Qwen models** support web search via `enable_search`: ```bash curl -X POST "https://api.mixroute.ai/v1/chat/completions" \ -H "Content-Type: application/json" \ -H "Authorization: Bearer sk-xxxxxxxxxx" \ -d '{ "model": "qwen3.7-max", "messages": [ {"role": "user", "content": "Hello"} ], "enable_search": true, "search_options": { "search_strategy": "standard", "forced_search": false } }' ``` ### File Input GPT models accept direct file content via the messages array: ```json { "role": "user", "content": [ {"type": "text", "text": "Analyze this document"}, {"type": "file", "file": {"url": "https://example.com/document.pdf"}} ] } ``` Supported file types: PDF, Word, Excel, images. ### Vision (Multimodal) ```python response = client.chat.completions.create( model="claude-opus-4-7", messages=[ { "role": "user", "content": [ {"type": "text", "text": "What's in this image?"}, { "type": "image_url", "image_url": {"url": "https://example.com/image.jpg"} } ] } ] ) ``` Base64 images also supported via `data:image/jpeg;base64,...` URI format. ### Response Format Non-streaming response: ```json { "id": "chatcmpl-xxx", "object": "chat.completion", "created": 1234567890, "model": "claude-opus-4-7", "choices": [ { "index": 0, "message": { "role": "assistant", "content": "Response content..." }, "finish_reason": "stop" } ], "usage": { "prompt_tokens": 25, "completion_tokens": 100, "total_tokens": 125 } } ``` ## Embeddings ```python response = client.embeddings.create( model="text-embedding-3-small", input="The quick brown fox jumps over the lazy dog" ) embedding = response.data[0].embedding print(f"Dimensions: {len(embedding)}") ``` Batch embeddings supported by passing an array to `input`. ## Image Generation ```python response = client.images.generate( model="gpt-image-2", prompt="A serene Japanese garden with cherry blossoms", size="1024x1024", quality="standard", n=1 ) image_url = response.data[0].url ``` For Google image models, use the canonical IDs `gemini-3-pro-image-preview` or `gemini-3.1-flash-image-preview`. For Qwen image models, use `qwen-image-2.0-pro`, `qwen-image-max`, or any of the `qwen-image-edit-*` variants. ## List Models ```python models = client.models.list() for model in models.data: print(f"{model.id} - {model.owned_by}") ``` ## Async Usage ```python import asyncio from openai import AsyncOpenAI client = AsyncOpenAI( api_key="sk-xxxxxxxxxx", base_url="https://api.mixroute.ai/v1" ) async def main(): response = await client.chat.completions.create( model="claude-opus-4-7", messages=[{"role": "user", "content": "Hello!"}] ) print(response.choices[0].message.content) asyncio.run(main()) ``` ## Environment Variables ```bash export OPENAI_API_KEY="sk-xxxxxxxxxx" export OPENAI_BASE_URL="https://api.mixroute.ai/v1" ``` The OpenAI SDK reads these automatically. --- # Error Codes | Status Code | Description | Resolution | | ----------- | --------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | 400 | Bad Request | Mismatch in request parameters. Verify your request body and structure. | | 401 | Invalid Token | Mismatch between API token and endpoint URL. Verify your API key is correctly set in the `Authorization: Bearer` header. | | 403 | Token Group Disabled | Token quota exhausted. (Token quota and account quota are separate.) Go to **Token Management → Edit → Enable "Unlimited Quota"**. Changes propagate in about 2 minutes. Alternatively, create a new token with Unlimited Quota enabled. | | 404 | Endpoint Not Found | API endpoint URL is incorrect, or the model ID does not match a model in the catalog. Verify against the live `/v1/models` endpoint. | | 429 | Rate Limit Exceeded | Model has reached its traffic limit. TPM (Tokens Per Minute) is saturated due to high usage. Retry after a short wait. If the issue persists, contact support with the full model name. Enterprise Reserve Capacity customers do not hit standard rate limits on the covered models. | | 500 | Internal Server Error | If error persists after multiple retries, contact support. | | 503 | Model Unavailable | Verify the model name on [console.mixroute.ai/models](https://console.mixroute.ai/models). If the name is correct but the error persists, provide the model name to support. | | 504 | Gateway Timeout | Upstream provider failed to respond in time. Retry the request. | ## Python Error Handling Pattern ```python from openai import OpenAI, APIError, RateLimitError, APIConnectionError client = OpenAI( api_key="sk-xxxxxxxxxx", base_url="https://api.mixroute.ai/v1" ) try: response = client.chat.completions.create( model="claude-opus-4-7", messages=[{"role": "user", "content": "Hello"}] ) except RateLimitError: print("Rate limit exceeded. Please retry later.") except APIConnectionError: print("Connection error. Check your network.") except APIError as e: print(f"API error: {e.message}") ``` --- # Billing and Usage Tracking ## Billing Model MixRoute operates on a **Pay-as-you-go** model. 1. **Real-time Deduction**: Fees are deducted from your balance immediately after each API request. 2. **No Minimum Spend**: There is no minimum consumption requirement. You only pay for what you use. 3. **No Expiration**: Your balance never expires. 4. **Pass-through Provider Rates**: Different models have different input and output rates. No platform markup on top of provider pricing. For per-model rates and the full live catalog, see [console.mixroute.ai/models](https://console.mixroute.ai/models). For deposit tiers, bonuses, and current promotions, see [mixroute.ai/pricing](https://mixroute.ai/pricing/). ## Tracking Usage 1. **Console**: Log in to the console and navigate to **Log Management → Usage Logs** to view historical records. 2. **API Response**: The `usage` field in every API response returns real-time token consumption (`prompt_tokens`, `completion_tokens`, `total_tokens`) for that request. ## Low Balance Alerts Set up low balance alerts in **Wallet Management → Alert Settings** to receive notifications before your balance runs out. --- # Frequently Asked Questions ## How do I switch between models? Change the `model` string in your API call. No SDK swap, no auth flow change, no separate billing relationship. The OpenAI SDK pointed at `https://api.mixroute.ai/v1` handles every supported model with the same code. ```python # Use Claude Opus 4.7 response = client.chat.completions.create( model="claude-opus-4-7", messages=[{"role": "user", "content": "Hello"}] ) # Switch to GPT-5.5 response = client.chat.completions.create( model="gpt-5.5", messages=[{"role": "user", "content": "Hello"}] ) # Switch to DeepSeek V4 Pro response = client.chat.completions.create( model="deepseek-v4-pro", messages=[{"role": "user", "content": "Hello"}] ) # Switch to Grok 4.20 (reasoning variant is the recommended default) response = client.chat.completions.create( model="grok-4.20-beta-0309-reasoning", messages=[{"role": "user", "content": "Hello"}] ) ``` ## Why is my API Key invalid? Three common causes for a 401 error: 1. **Typo or extra whitespace** in the key. Copy directly from the console. 2. **Wrong header format**. Must be `Authorization: Bearer sk-xxxxxxxxxx`. 3. **Mismatched base URL**. The key only works against `https://api.mixroute.ai/v1`. Create a fresh key in **Token Management** if the issue persists. ## Do I need a proxy to use the API? No. `api.mixroute.ai` is accessible globally without a proxy. ## What programming languages are supported? Any language with an HTTP client. Officially tested SDK paths: - Python (`openai` package) - Node.js / TypeScript (`openai` package) - Go (`github.com/sashabaranov/go-openai` or `github.com/openai/openai-go`) - Java (`theokanning/openai-java`) - PHP (Guzzle HTTP client) - Ruby (`ruby-openai` gem) - C# / .NET (OpenAI .NET SDK) - Rust (`async-openai` crate) The OpenAI API format is the standard. If your language can speak HTTPS and JSON, it can use MixRoute. ## How do I set up low balance alerts? In the console, go to **Wallet Management → Alert Settings**. Set the balance threshold and notification email. Alerts trigger when your balance drops below the set value. ## What model should I use? For general chat and content creation: GPT-5.5, Claude Sonnet 4.6, Gemini 3.1 Pro Preview, GLM-4.5 Air. For complex reasoning and data analytics: Claude Opus 4.7, GPT-5.5, GPT-5.4 Pro, Gemini 3.1 Pro Preview, DeepSeek V4 Pro, `grok-4.20-beta-0309-reasoning`. For high-throughput batch processing: Claude Haiku 4.5, Gemini 3.5 Flash, Grok 4 Fast variants, GLM-4 Flash. For code generation: Claude Opus 4.7, GPT-5.5, the Codex variants of GPT-5.x, DeepSeek V4 Pro, Qwen 3.7 Max, Grok Code Fast. For image generation: GPT-image-2, Gemini 3 Pro Image Preview, Qwen-image-2.0-pro, Qwen-image-max. For image editing: Qwen-image-edit-max, Qwen-image-edit-plus. For long-context workloads: Flagship models from Anthropic, Google, DeepSeek, and Alibaba all support extended contexts. Check current context window per model at [console.mixroute.ai/models](https://console.mixroute.ai/models). For reasoning tasks: o-series (OpenAI), DeepSeek R1, Claude `-thinking` variants, Grok 3 mini with `reasoning_effort=high`. ## What's the difference between models with `-thinking` suffix? Anthropic Claude models with `-thinking` (e.g., `claude-opus-4-7`, `claude-sonnet-4-6-thinking`) expose extended thinking, which surfaces the model's reasoning trace in a separate field. Comparable to DeepSeek R1's `reasoning_content` field. ## What is content safety policy? MixRoute enforces compliance with provider terms of service. Content that violates upstream provider policies will be blocked at the provider layer. MixRoute does not perform additional content moderation beyond what providers already enforce. Repeated terms-of-service violations may result in account suspension. --- # Integration Guides MixRoute works with every major AI tooling ecosystem. The integration pattern is consistent: change the API base URL to `https://api.mixroute.ai/v1` and use your MixRoute API key. ## Supported Integrations | Tool | Type | Integration Pattern | | --------------------- | --------------------------- | ------------------------------------------------------------------------------------------------------------------------ | | Cursor | Code editor | Settings → Models → Custom OpenAI Base URL. Point to `https://api.mixroute.ai/v1` with your MixRoute key. | | Claude Code | CLI for Claude | Set `ANTHROPIC_BASE_URL=https://api.mixroute.ai/v1` and `ANTHROPIC_API_KEY=sk-...`. Uses the `/v1/messages` endpoint. | | Claude Code for VS Code | VS Code plugin | Configure base URL in extension settings. | | Cherry Studio | Desktop AI chat client | Add MixRoute as a custom OpenAI provider in settings. | | Chatbox | Desktop and mobile AI chat | Add as a custom OpenAI provider. | | Cline | VS Code AI assistant | Configure base URL in extension settings. | | LangChain | LLM framework | `ChatOpenAI(base_url="https://api.mixroute.ai/v1", api_key="sk-...", model="claude-opus-4-7")`. | | Dify | LLMOps platform | Add MixRoute as a custom OpenAI-compatible provider. | | n8n | Workflow automation | Use the OpenAI node with custom base URL. | | Immersive Translate | Browser translation | Use the OpenAI-compatible model option, set custom API URL. | | Kilo Code | VS Code coding plugin | Configure base URL in extension settings. | | LobeHub | AI chat platform | Add MixRoute as a custom provider. | | OpenCode | Multi-model coding assistant | Configure MixRoute API key and base URL. | | OpenClaw | Self-hosted AI assistant | Connect via MixRoute API key. | Detailed step-by-step setup for each tool is available at [docs.mixroute.ai](https://docs.mixroute.ai). ## Generic Integration Pattern For any tool that accepts an OpenAI-compatible base URL: 1. Find the API settings section. 2. Set the base URL or API endpoint to `https://api.mixroute.ai/v1`. 3. Set the API key to your MixRoute token (`sk-...`). 4. Set the model to any canonical MixRoute model identifier (see Complete Model Catalog above). The OpenAI SDK compatibility means tools written for OpenAI work with MixRoute by changing two values. --- # Use Cases - **Multi-model agent and tool-use workloads**: Route per step. Cost-sensitive loops on smaller or open-weight models, final review passes on flagship models, all from the same SDK call with only the model name changing. - **Coding agents**: Switch between Codex variants (GPT-5.x Codex family), Claude Opus 4.7, DeepSeek V4 Pro, and Qwen 3.7 Max based on task type, language, and budget. - **Long-context document analysis**: Flagship models from multiple providers support extended-context workloads. Switch between them based on availability and provider strengths. - **RAG pipelines**: OpenAI and Gemini embeddings plus generation models through the same key, plus optional document reranking via the `/v1/rerank` endpoint. - **Image generation across providers**: GPT image, Gemini image, Qwen image variants, and Qwen image editing variants all callable from one endpoint. - **High-volume image workloads**: Enterprise customers can secure Reserve Capacity on Google image models with committed TPM by time zone (see Reserve Capacity section above). - **Reasoning-heavy workflows**: o-series (OpenAI), DeepSeek R1, Claude `-thinking` variants, Grok 3 mini with `reasoning_effort=high` accessible through the same API. - **Crypto-native AI products**: USDT and USDC deposits with no KYC remove fiat-rail friction. - **Migration from OpenRouter, direct OpenAI, or direct Anthropic**: One-line base URL change. No SDK swap. No separate billing setup. - **Audio transcription**: Whisper-1 accessible through the standard endpoint. --- # Solutions by Audience - **For Developers**: One OpenAI-compatible API for every major model. One bill. One dashboard. Migration from a direct provider integration takes under 30 seconds. - **For Startups**: Zero markup means more runway. The Growth tier starts at $1,000 deposits with 3% bonus. USDT and USDC deposits work for teams without a US bank account or credit card. - **For Production Teams**: One integration absorbs the complexity of multiple provider SDKs, multiple auth flows, multiple billing relationships. Automatic OpenAI-format compatibility across every model in the catalog. - **For Enterprises**: Single top-ups from $30,000 unlock the full enterprise package: Reserve Capacity for select high-demand image models, custom routing strategies, dedicated account manager, 24/7 support, contract-based onboarding. - **For Agent Builders**: Multi-model routing per request through a single integration. Switch models per task without rewriting the calling code. - **For Image-Heavy Workloads**: Multiple image generation providers (OpenAI, Google, Qwen) accessible from one endpoint. Reserve Capacity available for Google image models at enterprise tier. --- # Languages MixRoute is available in 8 languages with full website localization. - [English](https://mixroute.ai/) - [简体中文 Simplified Chinese](https://mixroute.ai/zh-hans/) - [繁體中文 Traditional Chinese](https://mixroute.ai/zh-hant/) - [Français](https://mixroute.ai/fr/) - [Deutsch](https://mixroute.ai/de/) - [日本語](https://mixroute.ai/ja/) - [Русский](https://mixroute.ai/ru/) - [Tiếng Việt](https://mixroute.ai/vi/) --- # Resources - **Website**: [mixroute.ai](https://mixroute.ai/) - **Console**: [console.mixroute.ai](https://console.mixroute.ai/) - **Live Model Catalog**: [console.mixroute.ai/models](https://console.mixroute.ai/models) - **Documentation**: [docs.mixroute.ai](https://docs.mixroute.ai/) - **Pricing**: [mixroute.ai/pricing](https://mixroute.ai/pricing/) - **Reserved Capacity**: [mixroute.ai/reserved](https://mixroute.ai/reserved/) - **Blog**: [mixroute.ai/blog](https://mixroute.ai/blog/) - **FAQ**: [mixroute.ai/faq](https://mixroute.ai/faq/) - **Talk to Sales**: [mixroute.ai/talk-to-sales](https://mixroute.ai/talk-to-sales/) - **Support Email**: service@mixroute.ai --- # Legal - **Terms of Service**: [docs.mixroute.ai/en/service-term](https://docs.mixroute.ai/en/service-term) - **Privacy Policy**: [docs.mixroute.ai/en/privacy](https://docs.mixroute.ai/en/privacy)