The complete OpenAI-compatible API surface, governed.
Chat, embeddings, images, audio, batches, rerank, and moderations — all through one OpenAI-compatible base URL. Drop-in migration: change one environment variable and your existing SDK code works immediately.
Every endpoint you use today, governed by ManyLayers.
Chat & Completions
Full /v1/chat/completions and /v1/completions compatibility. Streaming, function calling, tool use, vision, and structured outputs — all forwarded faithfully to the underlying provider with zero format translation required by your client.
Embeddings
OpenAI-compatible /v1/embeddings routing to OpenAI, Cohere, Vertex AI, and self-hosted models including BGE-M3 and E5. Swap embedding providers without application changes.
Images
Generation, edits, and variations via /v1/images/generations, /v1/images/edits, and /v1/images/variations. Routes to DALL-E, Vertex Imagen, Stable Diffusion XL, and other configured providers.
Audio: Transcriptions & Translations
/v1/audio/transcriptions and /v1/audio/translations compatible with Whisper API and self-hosted Whisper models. PII scrubbing applied to transcript output before returning to caller.
Audio: Speech Synthesis
/v1/audio/speech compatible with OpenAI TTS and other configured synthesis providers. MP3, Opus, AAC, and FLAC output formats. Per-team character limits apply.
Batches, Rerank & Moderations
Batch inference jobs via /v1/batches for high-volume async workloads. Reranking endpoint for retrieval pipelines. Content moderation via /v1/moderations. All pass through governance and audit.
Drop-in migration in three steps.
Change your base URL
Set OPENAI_BASE_URL to your ManyLayers gateway. Any client using the OpenAI SDK — Python, Node.js, Go, Java — begins routing through the gateway immediately.
Use the full API surface
Chat, embeddings, images, audio, batches, rerank, and moderations all available at the same base URL. Use whichever endpoints your application already calls.
Governance applies automatically
Budgets, guardrails, routing rules, and audit logging apply to every endpoint and every request with no additional client configuration.
Governance applied uniformly across your entire API surface.
Most governance solutions cover chat but leave embeddings, images, and audio ungoverned. ManyLayers applies the same budget enforcement, PII firewall, guardrails, and audit logging to every endpoint — closing the gaps that compliance and finance teams always discover too late.
- Works with openai-python, openai-node, openai-go, and any OpenAI-compatible client
- Fine-tuning job submission and status via /v1/fine_tuning compatible endpoints
- Batches endpoint enables async inference for large document processing workloads
- Rerank endpoint improves retrieval quality for RAG pipelines without custom code
- All endpoints emit Prometheus metrics and write to the shared audit log
“We migrated five services — chat, embeddings, image generation, transcription, and batch jobs — to ManyLayers in a single afternoon. One base URL change per service. Everything just worked, and we had unified cost tracking the same day.”
Staff Engineer — Enterprise Software Platform