Sanva AI Gateway · 100% OpenAI-Compatible Format

All BytePlus Video & Image Models, DeepSeek Flash v4.1 & Frontier LLM Tokens

High-concurrency, pay-as-you-go API token relay. 100% compatible with the OpenAI API format—connect all BytePlus visual models (Seedance 2.5 / 2.0 video, Seedream 5.5 / 5.0 Pro / 4.5 image), DeepSeek Flash v4.1, and OpenAI, Claude, and Gemini with a single key.

Complete Model Catalog

All BytePlus Visual Models & Frontier Reasoning in One OpenAI-Compatible Endpoint

Seedance 2.5 · 2.0 Pro · 1.5 Lite

All BytePlus Video Models (Seedance 2.5 / 2.0 Pro / 1.5 Lite)

Full access to the entire BytePlus ModelArk video generation lineup with 4K/1080p storytelling, multi-reference control, and physics-aware motion.

  • Complete lineup: Seedance 2.5, Seedance 2.0 Pro, Seedance 1.5 Lite, and PixelDance
  • Text-to-Video (T2V), first/last-frame Image-to-Video (I2V), multimodal references, and video extension
  • High-throughput async task queue and webhook callbacks for commercial video pipelines
Seedream 5.5 · 5.0 Pro · 4.5 · SeedEdit

All BytePlus Image Models (Seedream 5.5 / 5.0 Pro / Flash / 4.5)

Complete BytePlus image generation, multi-image blending, and inpainting models with crisp lighting and bilingual typography.

  • Full access to Seedream 5.5, Seedream 5.0 Pro / Flash, Seedream 4.5 / 3.0, and SeedEdit
  • Up to 2048x2048 (4MP) output, style reference locking, and character consistency
  • 100% compatible with OpenAI Images API format for instant drop-in integration
DeepSeek Flash v4.1 · GPT-4o · Claude 3.7

DeepSeek Flash v4.1 & OpenAI / Claude / Gemini Token Relay

Featuring ultra-fast DeepSeek Flash v4.1 alongside the world’s top reasoning and coding models.

  • DeepSeek Flash v4.1, DeepSeek R1 / V3 for ultra-low-latency reasoning and agent workflows
  • Direct enterprise routing for GPT-4o / o3, Claude 3.7 Sonnet / Opus, and Gemini 2.5 Pro / Flash
  • 100% OpenAI Chat Completions compatible with tool calling, JSON mode, and SSE streaming
Seed Audio · Embeddings · Rerank

Seed Audio, Embeddings, Reranking & Speech Models

Production endpoints for film-grade audio synthesis, speech transcription, and vector search.

  • BytePlus Seed Audio 1.0 generation and low-latency Whisper speech-to-text
  • text-embedding-3-small / large, BGE-M3, and high-precision rerankers
  • Load-balanced enterprise pools that eliminate single-account 429 rate limits
100% OpenAI Format Compatible

Zero Code Rewrite. Swap Base URL & Ship Immediately.

Standardized on the OpenAI API specification. Plug directly into the official OpenAI SDK, Cursor, Dify, Coze, n8n, LangChain, or your custom backend.

Pay-As-You-Go & Volume Discounts

No monthly minimums. Transparent per-token and per-generation billing with wholesale tiers.

100% OpenAI Format Compatible

Unified /v1/chat/completions, /v1/images/generations, and /v1/videos/generations endpoints.

Enterprise RPM & 99.9% SLA

Multi-region load balancing, automatic failover, and real-time usage dashboard.

// 100% OpenAI 兼容格式调用 BytePlus 全系生视频 (Seedance 2.5 / 2.0 Pro / 1.5 Lite)
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://ai-api.sanva.top/v1",
  apiKey: process.env.SANVA_API_KEY,
});

const videoTask = await fetch("https://ai-api.sanva.top/v1/videos/generations", {
  method: "POST",
  headers: {
    "Authorization": `Bearer ${process.env.SANVA_API_KEY}`,
    "Content-Type": "application/json"
  },
  body: JSON.stringify({
    model: "byteplus-seedance-2.5", // 支持 seedance-2.5 / seedance-2.0-pro / seedance-1.5-lite
    prompt: "Cinematic 4K shot, neon cyber street in rain, smooth camera tracking, 60fps",
    resolution: "1080p",
    duration: 10
  })
});