v2.21.12

What's Changed

• fix(qwen3-vl): invalid incremental output by @kevinlin09 in #128
• fix(model): invalid incremental output by @kevinlin09 in #129
• fix(doc): java doc eror due to comment by @kevinlin09 in #130

Full Changelog: v2.21.11...v2.21.12

v2.21.11

What's Changed

• fix(assistant_api): high latency of first calling by @kevinlin09 in #127

Full Changelog: v2.21.10...v2.21.11

v2.21.10

What's Changed

• feat(model/vl): support thingking mode by @kevinlin09 in #124
• fix(model/chat): invalid increamental_output by @kevinlin09 in #125

Full Changelog: v2.21.9...v2.21.10

v2.21.9

What's Changed

• feat(model/rerank): support instruct by @kevinlin09 in #116
• feat(model/qwen-tts): add usage.characters and voice names by @songguocola in #118
• feat(model/qwen-tts):remove unused voice name by @songguocola in #119
• feat(model/qwen-tts): add param language_type by @songguocola in #122

Full Changelog: v2.21.8...v2.21.9

v2.21.8

• feat(model/ocr): support new task type

v2.21.7

• feat(model/qwen-asr,model/cosyvoice-v3): support more parameters
• feat(tingwu agent): add realtime api

v2.21.6

• support qwen-image parameters
• support kv-cahce parameters
• support sync call for ImageSynthesis

v2.21.4

1. support qwen-vl model input paramter response_format

v2.21.3

1. support text-rerank model
2. fix MultiModalDialog

v2.21.2

1. fix multimodal api