v3.2.0

TL;DR — GPT-6 Astra and Gemini 3.8 Flash are in the model picker. Astra is not the default: it costs twice what GPT-5.6 Sol does and is still rolling out by organization.

Added

• GPT-6 Astra (gpt-6-astra) — OpenAI's frontier model, released 3
  September. 1M context, 128K output, and /thinking works on it.

  Not the default, on purpose. It is $10/$50 per million tokens against 5.6
  Sol's $5/$30 — twice the price — and OpenAI is still rolling it out
  organization by organization, so a key that works today may not have it.
  GPT-5.6 Sol stays the default; Astra is one /model away.

• Gemini 3.8 Flash (gemini-3.8-flash) — Google's newest Flash, released 2
  September, at the same price as 3.7 Flash ($0.75/$3.75 promotional). 1M
  context, 64K output, tuned for long-horizon agentic work. If you are on 3.7
  Flash there is no cost reason not to move.

Existing model ids are untouched. New ones are added beside, never in place of —
dropping an id does not leave a pinned config on the older model, it drops the
lookup and lands that config on another provider's default.

Fixed

• /thinking would have been hidden on GPT-6. The control was gated on a
  gpt-5 prefix, so a reasoning model would have arrived without it and with no
  error to explain why.
• Gemini 3.8 would have returned 400 on every call. Google removed
  temperature, top_p and top_k in the 3.7 generation and 3.8 is built on 3.7, so
  it inherits the removal — the parameters are now withheld for it too.

Cost estimates use the promotional Gemini rate that runs to 2026-12-31. The rise
scheduled for 2027-01-01 is deliberately not written ahead of time: doing exactly
that for Sonnet 5 over-reported every run by 50% until it was caught by hand.
Cached reads on GPT-6 Astra are $1.00/M — 0.1x input, which the default rate
already applies.