• DeepL uses quality_optimized with loco_deepl_model filter
• DeepL beta languages enabled when next-gen model is used
• http_request_timeout filter returns maximum value
• Splitting chat completion jobs into maximum 5KB batches
• Added support for other chat completion APIs via vendor config field
• Fixed single colon usage bug during chat completions
• Autoloader no longer throws when Loco_ class not found
• Refreshing Ajax nonces on WordPress heartbeat
• Bumped WordPress compatibility to 6.9