Tinker vs Vertex AI Gemini tuning

Tinker

Full control of the training loop with the GPUs abstracted away, plus recipes for SFT, DPO, RL and distillation. LoRA only; no full-parameter training.

Vertex AI Gemini tuning

Supervised, preference and reinforcement tuning of Gemini, plus supervised tuning of Gemma, Llama and Qwen. No weight export. The tuned model exists only as a Google Cloud endpoint.

Tinker

1. Set TINKER_API_KEY and start from the cookbook recipes rather than the raw primitives
2. Read the 'Avoid Client-Side Timeouts and Retries' guide before wrapping sampling calls in your own retries; the SDK already retries sampling with stable request IDs
3. Save intermediate checkpoints with a TTL between 1 hour and 10 years; storage bills at $0.10 a GB-month until they expire
4. Read models.json for current prices before a run; sampling tokens cost more than training tokens on the open models
5. Check the model deprecations page before pinning a base model; 18 were retired on 2026-06-12

Vertex AI Gemini tuning

1. Use client.tunings.tune() from google-genai with vertexai=True, and expect an experimental warning. Tuning isn't available on the Gemini Developer API
2. Add .md.txt to any docs.cloud.google.com URL to read the page as Markdown
3. Tune Gemini 3.5 Flash or 3.1 Flash-Lite. The 2.5 models retire on 2026-10-20
4. List jobs with a filter before re-sending a create after a timeout. There's no request ID to deduplicate it
5. Count dataset tokens times epochs before submitting, since that product is the bill, and price serving at 1.5x base for Gemini 3 tunes