AI Today: Models, Plans & Changes Worth Checking

WISECHECKER / AI CHECK

Make the next AI decision clearer.

A source-linked briefing on AI access, pricing and changes. Compare ChatGPT, Claude and Gemini, or check Codex and Claude Code before you choose.

Edition reviewed 2026-10-22 · Individual plans · English edition

Check my options How we verify information
01 / THE BRIEFING

Changes worth checking

Source-checked updates, not an unattended news feed. Each item links to its evidence and tells you what to check next. The source date is not a promise that your account has received a rollout.

Provider announcement

GPT-5.5: Codex retirement scope and timing

OpenAI specifies October 14, 2026 at 19:00 UTC for retirement from Codex with ChatGPT sign-in. API-key Codex sessions and the OpenAI API are outside this notice.

What to do: Replace saved gpt-5.5 selections with a model available to your plan and client. Check your account; the published date is not an account-level test.

Announced effective date: 2026-10-14. The date alone does not confirm completion or availability in your account.

Source checked · Checked 2026-10-22T03:18:38Z

OpenAI: model access and retirement notes

Provider announcement

Claude Max: API credits have separate eligibility rules

Max 5x and 20x offer $100 and $200 in monthly API credits respectively. They do not cover interactive Claude Code or subscription extra usage, and do not reset plan limits.

What to do: An eligible active plan needs seven days before claiming. Check Console-linking roles and eligible API/SDK uses; unused credits expire each billing cycle. Do not subtract credits from the fee.

Source checked · Checked 2026-10-22T03:18:38Z

Claude: monthly API-credit eligibility

02 / PICK YOUR NEXT STEP

Three questions. Clearer next steps.

ACCESS

What does my plan include?

Compare the consumer service with the coding agent and the developer API separately. A familiar provider name does not mean one shared bill.

Compare plans and access →
USAGE

Did my limit reset?

Start with your account usage display. A regular reset, a provider-wide change and a claimable reward are three different events.

Understand Codex resets →
WORKFLOW

Which coding agent fits?

Check how you sign in, where the work runs and how you verify the result. Then test both tools on a task you can judge.

Compare Codex and Claude Code →

Model names need context

A model appearing on a provider page is not proof of access on every plan, in chat, in a coding agent or through an API. These are selected access notes, not a leaderboard or a complete model catalog.

ChatGPT

GPT-6.1 Sol / GPT-6 Luna

GPT-6.1 Sol is rolling out to paid plans, excluding Free/Go at launch. Free/Go have rollout-dependent Luna access in the desktop app.

These model notes concern Work and Codex, not ordinary Chat. Availability still depends on your plan, client and account.

Source checked

OpenAI: model access and retirement notes

Claude

Fable 5.1 / Opus 5.5 / Sonnet 5.5 / Haiku 5.5

The plan table lists Sonnet/Haiku on Free, Opus on Pro/Max, Fable via usage credits on Pro and within 50% of weekly limits on Max.

The names above are the listed model versions; the plan table describes model families. Verify the actual version and surface in your account.

Source checked

Claude: individual pricing

Gemini

Gemini 4 Argon

Listed on the current US Google AI Pro and Ultra plan cards.

An included model name does not establish a fixed number of prompts or identical feature access across countries.

Source checked

Google: Gemini subscriptions (US / USD)

Evidence and update policy

This independent reference is checked against provider documentation during editorial preparation, with software assistance. It is not an account-level test or a head-to-head benchmark. Prices, access notes and updates share one catalog, so a reviewed correction can flow to all five pages. Fetching a page or saving an edit does not, by itself, renew its verification date.

Each record has an explicit review deadline, no more than seven days after its check. Overdue records are labeled and excluded from price-based recommendations. Offers with an explicit expiry stop displaying as active after that time. Missing evidence means “not verified,” not “free,” “unavailable” or “no incident.”

Official sources establish documented terms; your provider account and local checkout establish your actual access and charge. We do not inspect either. There is no universal model ranking, automated publication of scraped claims or promise of continuous human coverage.

Official sources and review dates