readme: document HTTP/2 connection pooling
This commit is contained in:
parent
209e225081
commit
bea543aa28
|
|
@ -47,6 +47,7 @@ Monorepo with Git submodules. Each submodule has its own Go module (except `el`
|
|||
|
||||
- **Multi-backend routing** — Ollama (local), OpenAI-compatible, Anthropic, GitHub Copilot, Kiro; switch backends per-session
|
||||
- **Model listing and switching** — enumerate available models from any backend; switch models mid-session
|
||||
- **Connection pooling** — all sessions sharing the same API host multiplex over a single HTTP/2 connection; no redundant TLS handshakes
|
||||
- **24-hour model cache** — model lists are cached at the backend level to avoid repeated API calls
|
||||
|
||||
### Tools and sandboxing
|
||||
|
|
|
|||
Loading…
Reference in New Issue