Text
LLM Inference
4 models including Llama 4 Scout, Qwen3 32B, and more. Sub-second responses. Multi-model selection via input.model. Smart routing picks the fastest available backend.
Text, image, video, audio, vision, documents — 51 models behind a single REST endpoint. Sub-second LLM responses. Automatic failover across 5 backends. Pay per request with USDC or credit card.
Every AI capability you need, accessible through POST /run
4 models including Llama 4 Scout, Qwen3 32B, and more. Sub-second responses. Multi-model selection via input.model. Smart routing picks the fastest available backend.
FLUX Schnell/Dev/Pro, SD 3.5, Lightning. Video from text or image. ControlNet, inpainting, outpainting, AI portraits, stickers, and product ad generation.
Whisper STT in under 1 second. TTS with 40+ voices and languages. Expressive speech with sound effects. Music generation. Voice cloning. Speaker diarization.
Image captioning, visual Q&A, CLIP prompt interrogation, OCR text extraction, object segmentation, and NSFW content detection.
PDF/document parsing to structured markdown. Document reranking for RAG pipelines. Video enhancement up to 4K. Background removal, face restoration, and more.
Multi-backend failover with circuit breakers. If one backend is overloaded, we retry on the next. X-Priority: fast|cheap header for routing control. Idempotency keys for safe retries.
Pay only for what you use. Zero cost when idle. No hidden fees.
Volume discounts: 5% off at $100 spent, 10% at $500, 15% at $1,000+. Applied automatically. Cost estimator →
From zero to running AI jobs in under 2 minutes.
Create an account with your email. Get your API key instantly. No credit card needed.
Top up from $10 via credit card. Bonus credits on $25/$50/$100 packages. Or use USDC per-request.
Pick a service, send your input. We route to the best available backend automatically.
Poll status or receive via webhook. You pay only for actual compute time. Excess is refunded.
Built for autonomous AI agents and human developers alike.
For autonomous AI agents. Send USDC on Base, include tx proof as header. No account, no API key.
Top up your account with USDC on Base. Same credit packages as card, with only 0.5% fee.
Register once, buy prepaid credits with card. Volume discounts and spending limits.
GPU-Bridge is a unified AI API offering 30 services and 51 models across text, image, video, audio, vision, documents, and search. It routes jobs across 5 enterprise GPU backends with automatic failover, so you get results even when individual backends are overloaded. You pay only for actual compute time.
LLM inference (4 models, sub-second), image generation (FLUX, SD3.5, Lightning, ControlNet, inpainting), video generation and enhancement (up to 4K), speech-to-text (Whisper, under 1 second), TTS (40+ voices), music generation, voice cloning, embeddings, document reranking (Jina), OCR, PDF/document parsing, NSFW detection, image captioning, visual Q&A, face restoration, background removal, upscaling, sticker generation, portrait generation, and more. See /catalog for the full list.
Three options: (1) USDC direct on Base — for AI agents, no account needed, pay per request by sending USDC and including tx proof in the X-Payment header; (2) Crypto top-up — register an account, top up credits with USDC on Base via POST /account/topup-crypto (only 0.5% fee); (3) Credit card via Stripe — register with email, buy credits from $10 (bonus credits on $25/$50/$100 packages). Volume discounts of 5-15% apply automatically based on total spending.
GPU-Bridge routes every job across multiple independent GPU backends. If the primary backend fails or times out, your request is automatically retried on the next available backend. Circuit breakers prevent routing to degraded backends. If all backends fail, you receive a full refund. Zero manual intervention needed.
LLM inference completes in under 1 second. Image generation takes 2-15 seconds depending on model and resolution. Video generation takes 60-300 seconds. Speech-to-text processes faster than real-time. Most utility services (OCR, captioning, background removal) complete in 2-10 seconds.
Yes. Any AI agent with USDC on Base can call any GPU-Bridge endpoint without registration. Include an x402 payment header in the HTTP request and the payment settles instantly on-chain. No human intervention, no API key, no account required.
Yes. Every account has a configurable daily spending limit (default $50/day) to prevent runaway costs from automated workflows. Adjust it anytime via POST /account/spending-limit. Range: $1 to $10,000 per day.
Yes. GPU-Bridge is operated by Healthtech Capital LLC, a registered Wyoming LLC (30 N Gould St Ste R, Sheridan, WY 82801). Payments are processed through Stripe (PCI DSS compliant). All crypto addresses are screened for OFAC compliance. Infrastructure runs on dedicated servers in Ashburn, Virginia with enterprise-grade security.
Free account. No credit card to register. 30 services ready to go.