AI Potluck
Model components / Fine-tuning code

Vertex AI Tuning

Google Cloud

Managed supervised fine-tuning and preference tuning for Gemini 2.0/2.5 Flash and Pro inside Google Cloud Vertex AI, covers text, image, audio, and document tuning with as few as 100 labeled examples. Picked when teams need Gemini-tier quality customized to a domain and want GCP-native data residency, IAM, and Vertex pipeline integration. The only path to fine-tune Gemini weights.

Google Cloud Vertex AI tuning for Gemini (docs live June 2026). Methods: supervised fine-tuning and preference tuning, across text/document/image/audio/video and function-calling. Proprietary managed Google Cloud service.

Openness

1 high confidence
1.0
source
closed
training-code
closed
runs-on
Google-Cloud-Vertex-AI-only
license
Proprietary(managed GCP service)

Proprietary managed tuning service inside Vertex AI; no source, runs only on Google Cloud against closed Gemini models.

Adoption

not assessed

No disclosed usage figure for the Vertex Gemini tuning feature located this run. The Gemini app surface has 750M+ MAU but that is not the Vertex tuning service and cannot be attributed to this SKU; declining to assign a level rather than borrow the surface number.

Capability

4 medium confidence
4.0

Broad managed multimodal tuning (SFT + preference + function-calling, across modalities) over frontier closed Gemini models. Black-box (no scale/precision controls, no benchmark basis), so capped below the OSS scale definers; scored 4 on method+modality breadth, comparable to the OpenAI FT API.

Unchanged since 2026-06-24 (last edited, not re-checked)