Say you pay for ChatGPT Plus. You spend the afternoon having Codex refactor a project, then talk through a design with GPT-6 Astra in ChatGPT after dinner. Today those come out of two separate buckets. From December they won't. OpenAI's help center says Plus, Pro and Business will share one allowance across ChatGPT and Codex, measured in tokens, because it is bringing the agentic capabilities of Work into the main ChatGPT experience.
Tokens mean OpenAI stops counting how many messages you send and starts weighing how much work each one takes. More advanced models and complex tasks generally use more. Instant chat will not draw from the allowance at all. OpenAI expects most users will not need to wait for a reset more often than they do now. That is its forecast, and whether it holds depends on numbers nobody has seen yet.
The trouble is that the parts that matter most are still missing: the exact December date, how many tokens your plan will get, and whether today's five-hour and weekly windows survive. The best guide for now is how Work and Codex already work. Per OpenAI's GPT-6 Astra usage guide, the two already share one allowance, with chat counted separately. OpenAI says that allowance "does not correspond to a fixed number of messages or tasks", then gives ranges anyway: in a five-hour window, Plus and Business Standard users get roughly 5 to 45 local messages with GPT-6 Astra, or 350 to 3,000 with GPT-6 Luna, and OpenAI stresses these are estimates. A range that wide tells you message counts were never a good yardstick.
What December really adds is your non-Instant chat. It gets folded into the pool that Work and Codex already share. Today you need allowance left in both the five-hour and the weekly window to keep going, while Pro 100, Pro 200 and Pro 500 have no five-hour limit in Work and Codex but still have an overall allowance. Run out, and you can wait for the reset, upgrade, or buy credits to keep using eligible features. OpenAI points to Settings → Usage for remaining usage and reset times, so it is worth glancing at it this month to learn your own habits before the new numbers arrive.
Claude Max 20x is now Ultra, and the price hasn't moved
Anthropic's Ultra plan page says Max 20x is now called Ultra and existing subscribers move over automatically with no price change: $200 a month on the web, with mobile pricing possibly different by platform. Ultra gets 10 times Pro's usage every week, under one weekly limit for all models plus a separate weekly limit for Fable. The Max plan stays at $100 a month with five times Pro's usage every five hours and every week. The page gives no date for the rename and doesn't explain how the old "20x" relates to today's "10 times Pro every week".
The line heavy users will care about: verified Ultra accounts have no five-hour limit. Verification happens automatically based on account and payment history. Until then there is a temporary limit, lifted once you're verified, and verifying a phone number at checkout won't lift it. To check, open Settings > Usage. If there is no "Current session" bar, you're verified.
Anthropic hasn't moved to token-based allowances. Its usage limits page still talks about a "conversation budget" that depends on conversation length and complexity, the features you use, the model and the effort level, with higher effort using more tokens. claude.ai, Claude Code and Claude Desktop all count toward the same limit. Tokens affect your usage there, but the allowance itself isn't stated in tokens. For now, OpenAI is the one that has said it outright.
The free API credits won't stretch your chat or Codex time
Both companies' higher plans also come with monthly API credits (OpenAI help article, Claude Help Center), sized by tier and seats:
- ChatGPT Pro: the $100, $200 and $500 tiers get $100, $200 and $500 a month.
- ChatGPT Business: $20 per paid Standard seat and $100 per paid Premium seat, pooled for the workspace and capped at $500 a month, or $6,000 for annual billing.
- Claude Max and Ultra: $100 and $200 a month.
- Claude Team: $20 per Standard seat and $100 per Premium seat, pooled and capped at $500.
Both companies cap the credits at $500 a month. Only ChatGPT Pro 500, and Business or Team workspaces with enough seats, reach that.
The catch is where the money can go. OpenAI's credits apply only to the linked API organization: they do not raise ChatGPT or signed-in Codex limits, and they don't cover Codex used through API keys. Plus, free trials, Enterprise and Edu are not eligible, the paid subscription must have been active for seven days before you can claim, and whether you see the offer also depends on rollout cohort and country. Anthropic's credits don't cover interactive Claude Code, extra usage inside Claude, or Claude on Bedrock, Vertex AI or Foundry, and Free, Pro and Enterprise are not eligible. At both, unused credits expire at the end of the cycle and do not roll over.
None of these OpenAI or Anthropic documents mentions a price increase; what changes is how fast you spend the allowance. If you mostly fire off quick Instant questions, December barely touches you. If you lean on Codex for long jobs and strong models for long chats, those two habits will start competing for the same pool. Wait for OpenAI's per-plan token numbers before you think about switching plans.