ModelGate OSS v0.1.0

ModelGate OSS v0.1.0

ModelGate OSS v0.1.0 is an early release of a local-first LLM gateway. It is not yet recommended for critical production workloads without independent validation.

Included

• OpenAI Chat Completions and Responses forwarding
• Anthropic Messages forwarding
• Incremental streaming relay
• Local SQLite request metadata
• Token counts, latency, errors, and estimated costs
• Searchable local dashboard and analytics API
• Exact normalized repeated-request grouping
• Basic transparent heuristic security rules
• Metadata-only privacy default with optional local content capture
• Docker Compose, mocked tests, examples, and deterministic demo data

Known limitations

• Dashboard/API authentication is not built in; keep it local or deploy behind authenticated TLS termination.
• Pricing estimates come from a manually maintained dated registry and may differ from provider billing.
• Streaming token counts require usage events from the provider.
• Repetition detection does not perform semantic similarity.
• SQLite is intended for a single local instance, not multi-replica deployment.
• Provider compatibility focuses on the documented paths above and is not exhaustive.

Upgrade notes

This is the first release. APIs and the SQLite schema may change before a stable release. Back up local data before upgrading.