Skip to content

docs: GPT-6.1 Sol — $2.00 / $10.00, cached $0.10, 1M; counts 88 / 115 - #129

Merged
VickyXAI merged 1 commit into
mainfrom
docs/gpt-6-1-sol
Oct 9, 2026
Merged

VickyXAI merged 1 commit into
mainfrom
docs/gpt-6-1-sol

Conversation

@VickyXAI

@VickyXAI VickyXAI commented Oct 9, 2026

Copy link
Copy Markdown
Contributor

Adds openai/gpt-6.1-sol (OpenAI's 2026-09-29 refresh of GPT-6 Sol) to the public docs.

  • models.md — row in the GPT-6 table; intro now covers four models (same price as Sol, cached $0.10/M vs $0.20/M, 272K tier at $4.00/$15.00, no Flex yet); none/minimal raised to low on 6.1 Sol. Also fixes the stale "82 chat / LLM models" note to 88.
  • pricing.md — OpenAI table row ($2.00 / $10.00 / cached $0.10).
  • changelog.md — ## [2026-10-09] / ### Added — GPT-6.1 Sol.
  • Count sweep — visible chat models 87 → 88, total catalog 114 → 115 across 21 pages.
  • enterprise-api.md — long-context ladder count 25 → 26.

Note: the parent repo's src/lib/brand-numbers.docs.test.ts pins the long-context count at 25 (EXEMPT entry + .toBe(25)); it must be bumped to 26 alongside the submodule pointer. With that bump, all 11 tests pass against this branch.

🤖 Generated with Claude Code

openai/gpt-6.1-sol on the price sheet, the models reference (GPT-6 table and
intro, now four models) and the changelog. Same price as gpt-6-sol with half
its cached rate; 272K long-context tier at $4.00 / $15.00; no Flex yet.
Count sweep 87 -> 88 chat, 114 -> 115 total, the stale "82 chat / LLM models"
note on the models page, and the long-context ladder count 25 -> 26.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
@VickyXAI
VickyXAI merged commit ff684ad into main Oct 9, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant