Manifest vs TensorZero
Manifest
manifest.build
TensorZero
tensorzero.comSide by side
| Figure | Manifest | TensorZero |
|---|---|---|
| Price from | $19/mo | Free |
| Free plan | Yes | Yes |
| Free trial | — | — |
| Open source | Yes | — |
| Web | Yes | Yes |
| Linux | Yes | Yes |
| API | Yes | Yes |
| Self-hosted | Yes | Yes |
| Plans published | 4 | 1 |
| Spec score · LLM gateways | 9.4 (#1) | 8.6 (#7) |
| Spec score · LLM Gateway Software | 9.0 (#4) | 8.0 (#13) |
| Integrations | Manifest connects API keys, paid subscriptions, local models, and custom endpoints; named local integrations include Ollama, llama.cpp, and LM Studio. | Documented providers include Anthropic, AWS Bedrock, Azure, Google, OpenAI, OpenRouter, Together AI, vLLM, and OpenAI-compatible endpoints. |
| Observability | Usage views show spending, selected models, token consumption and failed requests for each harness. | TensorZero can store inference and feedback data for observability, optimization, evaluations, and experimentation in Postgres or ClickHouse. |
| Product | Manifest is an open-source LLM gateway that routes requests, provides fallbacks, repairs failing requests with Autofix, and tracks spending. | TensorZero is an open-source platform that combines an LLM gateway, observability, evaluation, optimization, and experimentation. |
| Security | Manifest states that data in transit uses TLS, passwords are hashed, and Gateway provider credentials are encrypted with AES-256-GCM. | TensorZero's privacy policy says it uses physical, technical, organizational, and administrative security measures to protect personal data. |
| Support | Free includes community support through Discord, Pro includes basic platform and billing support, and Enterprise includes dedicated Slack and email support. | The project README directs users to Slack or Discord and offers work teams a Slack or Teams channel by email. |
Plans
In their own words
Manifest
Manifest is an open-source LLM gateway for routing model requests, switching providers, repairing certain failures, and tracking inference spending. It provides one endpoint compatible with OpenAI and Anthropic models and subscriptions. When a provider goes down, fallback routing retries through another provider. Autofix repairs malformed requests and resends them when retries alone cannot resolve the error. Observability shows spending by harness, which models handled requests, and what failed. Manifest connects API keys, paid subscriptions, local models, and custom endpoints, including Ollama, llama.cpp, and LM Studio. The self-hosted gateway runs with Docker, works offline, requires no account, and supports amd64 and arm64. Self-hosted routing and request data remain on customer servers except documented telemetry and Autofix flows. The Cloud Free plan allows 10,000 routed requests per calendar month and cannot reach local models on the user's machine. Free includes unlimited self-hosted use; Pro costs $19 a month with unlimited Cloud requests and 365-day dashboard retention.
TensorZero
TensorZero is a free, open-source stack that brings together an LLM gateway, observability, optimization, evaluation, and experimentation. Its gateway provides a unified API for model providers and supports routing, retries, fallbacks, prompt templates, schemas, multi-step workflows, and traffic routing for A/B tests. Documented integrations include Anthropic, AWS Bedrock, Azure, Google, OpenAI, OpenRouter, Together AI, vLLM, xAI, and OpenAI-compatible endpoints. Inference and feedback data go to the database you configure, where they can be queried with SQL; feedback can be tied to individual inferences or episodes. The gateway and UI can run through Docker, Docker Compose, Kubernetes and Helm, or a source build. API key authentication is available for gateway endpoints other than status and health checks. The project is available on GitHub, but is no longer maintained; its repository is archived and read-only. TensorZero is self-hosted and licensed under Apache 2.0. It collects pseudonymous aggregate usage analytics, excluding application data and inference inputs and outputs, with an option to disable collection.