LiteLLM
Product status: Status not recorded
No lifecycle review is recorded for this listing. Its inclusion in the directory does not confirm current availability; check the vendor before relying on it.
No status check date recorded
A model-routing SDK and self-hosted AI gateway with virtual keys, budgets, fallbacks and request-cost records; Enterprise adds licensed features. No lifecycle review is recorded for this listing. Its inclusion in the directory does not confirm current availability; check the vendor before relying on it.
Product preview
Decision notes
Official-document research
Based on the cited documentation. We did not use the product, purchase a plan, or run quality, speed or reliability tests in this review.
Consider it for
- Platform teams able to manage providers, keys and logging
Choose another approach if
- Treating a proxy as a universal no-data-egress guarantee
Pricing and conditions
Documented terms
OSS self-hosting has no software license fee. Enterprise is an annual quote based on gateway request capacity, architecture and support, not a per-model-token fee; its 30-day trial is not an ongoing free plan. Docs describe an SSO exception for up to 5 users. Provider inference, compute, database/log storage and operation remain separate. Enterprise amount/currency/tax, region, overage, trial conditions and the actual license agreement were not established.
Pricing source read:
Pricing or terms sourceCode, model and service licensing
The repository's MIT grant applies outside enterprise/, with notice obligations; enterprise/ is expressly governed by another license. Its referenced license file was unavailable, so the exact commercial agreement was not verified. Code rights do not grant provider access, model weights or cleared outputs. Security docs' no-telemetry/no-vendor-storage statement is about self-hosting and LiteLLM servers. Configured LLMs and logging callbacks can receive content; disabling message logging still tracks metadata such as spend. Operator/provider training, retention, support access and contracts must be reviewed separately; no universal no-egress or zero-retention promise is inferred.
Basis and limits of the recorded score
Original 4.6/5 retained; the original scoring worksheet and hands-on evidence are unavailable. This documentation review did not re-score the product.
Overview
LiteLLM offers a Python SDK and a gateway that route requests to configured model providers. The gateway adds virtual keys, budgets, rate limits, fallbacks and request-cost records; MCP/A2A integration adds further configured routes.
Operate the proxy, logging destinations and providers as separate data processors. An OpenAI-format request does not make every model compatible, equally priced or covered by the same contract.
Sources
The factual claims in this review were read from the pages below. Each date is when an editor last opened the source, not when the product changed.
- Vendor site Checked 2026-10-15
- Vendor documentation Checked 2026-10-15
- Vendor pricing page Checked 2026-10-15
- Vendor documentation Checked 2026-10-15
- Vendor documentation Checked 2026-10-15
- Vendor documentation Checked 2026-10-15
- Vendor documentation Checked 2026-10-15
Pros / Cons
Pros
- Centralized keys and usage controls
- SDK and gateway deployment choices
Cons
- Proxy and logs need operation
- Enterprise and external inference cost separately
Who it's for
Organization-wide model access budgets
Routing several provider APIs
Bottom line
Budget Enterprise capacity separately from provider inference and review every callback.
Key Features
Tags
Frequently asked questions
Is LiteLLM currently available?
No lifecycle review is recorded for this listing. Its inclusion in the directory does not confirm current availability; check the vendor before relying on it.
What is LiteLLM?
LiteLLM is a AI Platforms tool developed by BerriAI. A model-routing SDK and self-hosted AI gateway with virtual keys, budgets, fallbacks and request-cost records; Enterprise adds licensed features.
What costs and terms should I check for LiteLLM?
OSS self-hosting has no software license fee. Enterprise is an annual quote based on gateway request capacity, architecture and support, not a per-model-token fee; its 30-day trial is not an ongoing free plan. Docs describe an SSO exception for up to 5 users. Provider inference, compute, database/log storage and operation remain separate. Enterprise amount/currency/tax, region, overage, trial conditions and the actual license agreement were not established.
Is LiteLLM open source?
The repository's MIT grant applies outside enterprise/, with notice obligations; enterprise/ is expressly governed by another license. Its referenced license file was unavailable, so the exact commercial agreement was not verified. Code rights do not grant provider access, model weights or cleared outputs. Security docs' no-telemetry/no-vendor-storage statement is about self-hosting and LiteLLM servers. Configured LLMs and logging callbacks can receive content; disabling message logging still tracks metadata such as spend. Operator/provider training, retention, support access and contracts must be reviewed separately; no universal no-egress or zero-retention promise is inferred.
What are the best alternatives to LiteLLM?
Notable AI Platforms alternatives include Hugging Face, Replicate, Ollama. Browse all 138 tools in our AI Platforms category for a full overview.
Was this listing useful?
Basic Info
- Pricing
- Freemium
OSS self-hosting has no software license fee. Enterprise is an annual quote based on gateway request capacity, architecture and support, not a per-model-token fee; its 30-day trial is not an ongoing free plan. Docs describe an SSO exception for up to 5 users. Provider inference, compute, database/log storage and operation remain separate. Enterprise amount/currency/tax, region, overage, trial conditions and the actual license agreement were not established.
- Editorial score
- ★ 4.6 / 5 How we score
- Platforms
- Python SDKSelf-hosted Gateway
- Last verified
- Not recorded
