OmniRoute Supported Providers for Image, Video, and Audio Generation

OmniRoute supports 7 image generation providers, 8 video generation providers, and 1 dedicated audio provider, with several providers offering multi-modal capabilities across image and video generation.

OmniRoute is an open-source routing layer that unifies access to hundreds of AI providers through a single OpenAI-compatible API. The project's provider registry maps each provider to specific media generation capabilities, enabling developers to switch between image, video, and audio models without changing integration code. This guide details exactly which providers OmniRoute supports for each modality, with direct references to the source implementation.

Image Generation Providers in OmniRoute

OmniRoute supports 7 dedicated image generation providers, each accessible through standard /v1/images/generations endpoints or provider-specific routes. According to docs/reference/PROVIDER_REFERENCE.md, the following providers expose image generation capabilities:

Provider	Alias	Source Location
black-forest-labs	bfl	line 126
fal-ai	fal	line 156
freepik	fpk	line 162
recraft	recraft	line 252
segmind	segmind	line 262
stability-ai	stability	line 267
topaz	topaz	line 277

The image generation routing logic is implemented in open-sse/handlers/imageGeneration.ts, which normalizes provider-specific request formats into the OpenAI images API specification. This handler supports both synchronous generation and SSE streaming for progress updates.

Example: Generate an Image

curl -X POST https://localhost:20128/v1/images/generations \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-image-2","prompt":"a futuristic city skyline"}'

Video Generation Providers in OmniRoute

OmniRoute supports 8 video generation providers, including model families like Runway Gen-3, Haiper, and Pollinations. The provider reference tags these entries with video capability markers. Notably, segmind appears in both image and video lists, demonstrating OmniRoute's multi-modal provider support.

Provider	Alias	Source Location
haiper	hp	line 183
leonardo	leo	line 202
minimax	minimax	line 209
novita	novita	line 225
pollinations	pol	line 244
runwayml	runway	line 256
segmind	segmind	line 262
together	together	line 275

The open-sse/handlers/videoGeneration.ts file implements routing for /v1/videos/generations and provider-specific video endpoints. Video generation typically requires extended timeout configurations due to longer processing times compared to image generation.

Example: Generate a Video

curl -X POST https://localhost:20128/v1/providers/runway/videos/generations \
  -H "Content-Type: application/json" \
  -d '{
        "model":"runway/gen-3",
        "prompt":"a mountain landscape turning into a time-lapse",
        "num_frames": 30,
        "frame_rate": 24
      }'

Audio Generation Providers in OmniRoute

OmniRoute's audio support focuses on text-to-speech (TTS) and speech-to-text (STT) capabilities. The provider reference currently lists 1 dedicated audio provider with native audio generation support:

Provider	Alias	Source Location
fishaudio	fishaudio	line 344

Additionally, OmniRoute proxies standard OpenAI audio endpoints (tts-1, tts-1-hd, whisper-1) through provider-agnostic handlers in open-sse/handlers/audioSpeech.ts. This enables TTS and STT functionality across any configured upstream provider that implements the OpenAI audio API.

The audio handler supports both synchronous speech synthesis and streaming audio generation for real-time applications.

Example: Text-to-Speech Synthesis

curl -X POST https://localhost:20128/v1/audio/speech \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/tts-1","input":"Hello, welcome to OmniRoute!","voice":"alloy"}' \
  --output hello.mp3

Example: Speech-to-Text Transcription

curl -X POST https://localhost:20128/v1/audio/transcriptions \
  -H "Content-Type: application/json" \
  -d '{"model":"deepgram/whisper-1","audio_url":"https://example.com/audio.wav"}'

Multi-Modal Provider Capabilities

Some OmniRoute providers support multiple generation modalities. The segmind provider is explicitly tagged for both image and video generation in the provider reference. This pattern allows developers to use a single provider configuration for diverse media tasks, reducing credential management overhead.

When configuring multi-modal providers, open-sse/config/providerRegistry.ts maps each model ID to its appropriate executor based on the requested endpoint path. The registry validates that the selected model supports the requested modality before routing the request.

Key Implementation Files

Understanding OmniRoute's provider support requires familiarity with these core files:

File	Purpose
docs/reference/PROVIDER_REFERENCE.md	Canonical list of all supported providers with capability tags
open-sse/handlers/imageGeneration.ts	Image generation request routing and normalization
open-sse/handlers/videoGeneration.ts	Video generation request handling
open-sse/handlers/audioSpeech.ts	TTS/STT endpoint implementation
open-sse/config/providerRegistry.ts	Model-to-executor mapping and validation

These files collectively implement OmniRoute's unified API abstraction, allowing consistent access to provider-specific image, video, and audio generation capabilities.

Summary

• Image generation: 7 providers including Black Forest Labs, Fal.ai, Freepik, Recraft, Segmind, Stability AI, and Topaz
• Video generation: 8 providers including Haiper, Leonardo, Minimax, Novita, Pollinations, RunwayML, Segmind, and Together
• Audio generation: Native support for FishAudio plus OpenAI-compatible TTS/STT proxying
• Multi-modal: Segmind provides both image and video generation from a single provider configuration
• Implementation: Core routing handled in open-sse/handlers/ with provider definitions in docs/reference/PROVIDER_REFERENCE.md

Frequently Asked Questions

How do I add a new image generation provider to OmniRoute?

Create an entry in docs/reference/PROVIDER_REFERENCE.md with the image capability tag, then implement a provider-specific executor in open-sse/handlers/imageGeneration.ts or extend the existing routing logic. Register the model mapping in open-sse/config/providerRegistry.ts to enable request routing.

Can I use the same API key for multiple modalities with a single provider?

Yes, if the provider supports multiple modalities. The segmind provider demonstrates this pattern—configure once in your OmniRoute settings and route image requests to /v1/images/generations and video requests to /v1/videos/generations using the same credentials.

Why doesn't OmniRoute list more dedicated audio providers?

OmniRoute prioritizes OpenAI-compatible audio proxying through open-sse/handlers/audioSpeech.ts, which enables TTS/STT with any upstream provider implementing the standard /v1/audio/speech and /v1/audio/transcriptions endpoints. The fishaudio entry represents a provider with native, non-standard audio APIs requiring custom integration.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}