The developer platform for instruction-tuning open models. Define behaviour, train, evaluate, and iterate — all from a single API.
# 1. Define behaviour
curl -X POST https://api.tunedtensor.com/v1/behavior-specs \
-H "Authorization: Bearer tt_your_api_key" \
-H "Content-Type: application/json" \
-d '{
"name": "Support Agent",
"system_prompt": "You are a helpful support agent...",
"guidelines": ["Be concise", "Show empathy"],
"examples": [
{"input": "How do I cancel?", "output": "I understand..."}
],
"base_model": "Qwen/Qwen2.5-7B-Instruct-Turbo"
}'
# 2. Start a run — compiles, augments, trains, evaluates
curl -X POST .../v1/behavior-specs/{id}/runs \
-d '{"augment": true, "hyperparameters": {"n_epochs": 4}}'Three steps from intent to a behaviour-shaped model.
Describe what the model should do: system prompt, guidelines, input/output examples, constraints.
One click compiles your spec, augments examples with AI, fine-tunes the model, and evaluates it.
Review per-example pass/fail results, catch regressions, refine your spec, and run again.
Explore the documentation to see how behaviour specs, runs, and evaluations work under the hood.