* added GPT-4 Turbo with Vision * improved Anthropic Claude reliability by adding tool use * fixed a bug with image input
1.1.2
accept valid environment variables for zeebe (underscore instead of hyphen)
1.1.1
fix bug when selecting visual document extraction model
1.1.0
* added Anthropic Claude 3 model options * added AWS and Azure model options for extract, translate, and OCR * added Amazon S3 and Azure Blob Storage input * moved local AI inference into separate image in [bpm-ai-inference](https://github.com/holunda-io/bpm-ai-inference) project
1.0
* camunda-8-connector-gpt -> bpm-ai-connectors-camunda-8 * rewrote connectors in python using pyzeebe, as all AI/ML code uses python anyway * added support for local AI models from huggingface * added OpenAI Whisper audio transcription (local or API) * added image/pdf OCR using tesseract (local) * made default docker image **way** smaller * added optional larger image for local AI models