gpt-5.6-sol retired: it now runs on gpt-6-sol in Formats

A Sume Format run that still sends model: "gpt-5.6-sol" keeps working: the retired id runs on gpt-6-sol. And if you omit model, the Format run uses gpt-6-sol by default, per the Create a run page.

What does the docs line say exactly?

The model field is the Agents catalog id for the LLM that orchestrates the run. The page says to omit it for the gpt-6-sol default, that a request for retired gpt-5.6-sol runs on gpt-6-sol, that it selects the orchestrator only (image, video and audio models are chosen by the Format's tools), that an id outside the catalog is 400 invalid_request, and that the receipt echoes the id that ran.

That last point is your proof. Read the receipt after a run and compare with what you sent.

How do I test the alias safely?

Send the same body twice, once with the old id and once with the new, using separate Idempotency-Key values and a low generation_spend_cap_usd. Compare the model on each receipt. The cap is up to $500 platform-wide, and a value of 0 is rejected.

curl -sS -X POST "https://api.sume.com/v1/formats/acme/live-commerce/runs" \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: alias-check-1" \
  -d '{"instruction":"Use the default script.","model":"gpt-5.6-sol","generation_spend_cap_usd":5}'

What is OpenAI's side of this?

OpenAI's GPT-6 Sol page, read 2026-10-02, lists text and image input, text output, a 1,050,000-token window, $2 input and $10 output per million tokens and a cutoff of Apr 20, 2026. It does not mention the older 5.6 id, so I make no claim about OpenAI's own retirement of it. The alias is Sume's behaviour.

What should I change in my code?

Update stored payloads and templates to gpt-6-sol or drop the field and take the default. An alias is a courtesy, not a contract; the docs do not promise how long it stays, so do not rely on it for new work.

Remember that the orchestrator choice does not change which image or video model renders. For that, read the GPT-6 text-only post.