Release 0.24.8

Released 2016-09-22.

• Fixed token counting for long prompts
• Added per-model rate limits
• Improved streaming reliability

Release 30885 · Release 30886 · Release 17078

Back to the changelog