Release 0.5.6

Released 2019-03-31.

• Added per-model rate limits
• Fixed token counting for long prompts
• Added support for custom upstream timeouts

Release 75655 · Release 75656 · Release 19448

Back to the changelog