Oracle Cloud Infrastructure Documentation

Compatible OpenAI Models

You can import compatible models from Hugging Face and OCI Object Storage buckets into OCI Generative AI, create endpoints for those models, and use them in the Generative AI service.

OpenAI Whisper

The OpenAI Whisper Large V3 Turbo model is optimized for automatic speech recognition and audio transcription workloads. This audio-to-text model is a fine-tuned version of a pruned Whisper Large V3 model, with fewer decoder layers for faster transcription with a minor quality tradeoff. The model supports multilingual transcription, language identification, and speech translation from supported languages into English text and is suited for latency-sensitive and high-throughput audio processing use cases. For more details, see OpenAI Whisper Large V3 Turbo in the Hugging Face documentation.

Compatible OpenAI Whisper Model
Hugging Face Model ID Model Capability Recommended Dedicated AI Cluster Unit Shapes
openai/whisper-large-v3-turbo AUDIO_TO_TEXT
  • H100_X1
  • A100_80G_X1
Important

  • While you can import any chat, embedding, (and fine-tuned) model validated through Open Model Engine (with vLLM or SGLang runtime), only models explicitly listed on this page have been assessed for this model family by Oracle against open-source model runtimes and tested on Oracle-supported GPU configurations. Notwithstanding the foregoing, Oracle is not responsible for any issues related to the performance, availability, operation, or security of Compatible Models. Unlisted models might have compatibility issues and we recommend that you test any unlisted model before production use. Learn about OCI Generative AI Imported Model Architecture.

  • For imported models, you can use the native context length specified by the model provider. However, the effective maximum context length is limited by the underlying hardware setup that you select for the hosting dedicated AI clusters in OCI Generative AI. To take full advantage of a model's native context length, you might need to provision more hardware resources.
  • Use the fine-tuned models only if they match the compatible base model's transformer version and have a parameter count within ±10% of the original.
  • For available hardware and steps on how to deploy the imported models, see Managing Imported Models.
  • If a recommended shape isn’t available in a region, select the closest available alternative. For example, if H100_X2 isn’t available but A100_80G_X2 is, select A100_80G_X2. If both H100 and A100 shapes are available, for better performance, select H100.