Neura Market
Manifest logo

Manifest

Free

An open-source LLM router that routes agent requests to the most cost-effective model, with usage limits and model benchmarking. [#opensource](https://github.com/mnfst/manifest)

FreeFree tier
Type
Open Source
Visit Website GitHub (40 stars)

About Manifest

Manifest is an open source LLM router that gives developers full control over their AI agents by routing requests to the most cost-effective model and provider. It connects to multiple provider types—API key, subscription (Claude Pro, ChatGPT Plus, Gemini Advanced), custom, and local models (Ollama, LM Studio, llama.cpp)—and allows users to define routing rules, fallbacks, and usage limits. The platform includes Auto-fix to repair failed requests automatically, real-time cost dashboards, and budget alerts. Available as a hosted cloud service (with a free tier of 10,000 routed requests/month) or as a self-hosted open source (MIT licensed) deployment. Manifest is built for personal AI agents, AI applications, AI workflows, and coding assistants, aiming to reduce AI inference costs without sacrificing quality.

Key Features

Route LLM calls to the most cost-effective model and provider
Connect multiple provider types: API keys, subscriptions, custom, local models
Define routing rules and automatic fallbacks
Auto-fix failing requests on the fly
Real-time dashboard for cost tracking and usage visualization
Budget alerts and notifications
Supports OpenAI-compatible and Anthropic-compatible endpoints
Supports local models via Ollama, LM Studio, and llama.cpp
Free tier with 10,000 routed requests per month
Self-hostable with Docker and Docker Compose (MIT license)

Pros & Cons

Pros
  • Open source and self-hostable, eliminating vendor lock-in
  • Supports both subscription and pay-per-token providers for flexibility
  • Simple three-step setup: plug agents, add providers, route and control
  • Auto-fix feature minimizes downtime from failed requests
  • Free tier available with generous limits for prototypes and small projects
  • Comprehensive dashboard for real-time cost monitoring and optimization
Cons
  • Free cloud plan limited to 10,000 requests per month; requests pause if exceeded
  • Self-hosting requires Docker knowledge and infrastructure management
  • Subscription provider support depends on providers' continued API access
  • Team management features only available in Enterprise plan

Best For

Reduce costs for personal AI agents like OpenClaw or HermesOptimize inference efficiency for AI applications in productionIntegrate with AI workflow tools like n8n and DifyUse coding assistants with preferred models and local models

FAQ

Can I use my existing Claude or ChatGPT subscription?
Yes, on every plan including Free. Manifest connects to subscription-based providers (Claude Pro, ChatGPT Plus, Gemini Advanced) so your agents use those quotas before falling back to pay-per-token APIs. Availability may change if a provider alters their terms.
What's the difference between Free Cloud and the open source version?
Free Cloud is hosted by Manifest—no server to manage. The open source version runs on your own infrastructure with Docker, giving you full data control and no usage limits, but requires more setup.
What happens when I hit 10,000 requests on the Free plan?
Requests beyond 10,000 are paused until the next billing cycle. You can upgrade to Pro at any time to remove the limit.
Can I switch from Free to Pro mid-month?
Yes. Upgrading takes effect immediately and your billing cycle starts on the day you upgrade.
What is the difference between Free and Pro?
Both plans include unlimited agents, all providers, Auto-fix, and budget alerts. Pro removes the 10,000 requests/month limit, extends dashboard retention to 365 days, and includes basic email support for invoices.
Is the self-hosted version still free?
Yes. Manifest is open source and the self-hosted Docker image is free with no restrictions. The pricing on the page applies only to the hosted cloud version.