Manifest LLM Gateway

Manifest LLM Gateway is an open source LLM gateway that routes AI agent requests to any provider and repairs broken calls on the fly.

Manifest is an open source LLM router that connects AI agents and workflows to any provider in seconds. It is built for developers and teams to manage inference consumption, track requests, and prevent API errors. The gateway features request autofixing, model fallbacks, full body logs, and parameter setup. Users can bring their own API keys, use subscriptions, or run local models through Ollama, LM Studio, or llama.cpp. It is available as a cloud version or self-hosted via Docker with an MIT license.

Manifest LLM Gateway at a glance

Best for: Software engineer
Runs on: Web, API, self-hosted
Pricing: Free plan; Pro plan available; Enterprise plan with custom pricing
Released: September 2022
Category: AI coding tool
Website: manifest.build

Key features

Autofixes broken requests on the fly
Provides model fallbacks to another provider
Logs complete request and response bodies
Supports local models and custom providers
Visualizes cost broken down by agent and key
Configures parameters visually per route
Self-hosts on local machines or infrastructure

Use it for

Manage inference consumption and track tokens
Route AI calls to local models or cloud providers
Prevent client errors in production workflows
Monitor spending by agent and provider

Alternatives to Manifest LLM Gateway

llm-d — Deploys production-grade LLM serving on existing hardware clusters.
AG-UI — Standardizes agent-user interactions and generative UI for applications.
GoRoute — Consumes 5.1MB of RAM with sub-10ms latency for coding agents.
agentgateway — It unifies service, LLM, and MCP traffic into one open-source binary.
LiteLLM — One OpenAI-format API for 100+ LLM providers
Mesh LLM — Run large local models across multiple machines through one OpenAI-compatible endpoint.
LLM Space — Inspect every model call and tool run inside the agent loop locally.
OpenAPPA — Tracks data flows algebraically to prevent prompt injection without breaking agents.
AgentGate — Blocks attacks in under 35ms with a local neural classification head.
LLM Workflow Engine — Manage and execute complex LLM workflows directly from the command line.
OpenLLM — Routes developer tools through multiple providers and local models with included inference.
WebLLM — Runs high-performance language models directly inside web browsers.

All Manifest LLM Gateway alternatives