Which AI price changes actually affect you?

1. Your existing model’s rate changes

This can affect the same workload without switching models, subject to effective dates and account terms. October found no list-price change among the 17 models and plans shared by both editions. That is a matched sample, not a statement about every AI product.

2. A cheaper successor becomes available

The October small-model median fell because the selected lineup changed. For OpenAI, GPT-5.6 Luna’s $0.45 blended rate was replaced in the basket by GPT-6 Luna at $0.20. At ten million tokens with the same 3:1 input/output mix, those rates imply $4.50 and $2.00: a $2.50 difference. Accessing that modeled difference requires using the successor, and equal token counts do not establish equal task quality.

Anthropic’s selected small-model rate moved from Claude Haiku 4.5 at $2.00 to Haiku 5.5 at $0.20. The equivalent ten-million-token calculation is $20 versus $2. The October Haiku 5.5 observation uses the short-context band; longer prompts have different rates. Test your task and context length before treating this as savings.

3. Subscription tiers change

A stable median can hide a different product mix. October’s top-plan median remained $200, while OpenAI’s selected highest tier became the $500 ChatGPT Pro option. That is a $300 difference between selected tiers, not evidence that existing $200 subscribers were moved or charged more. Compare entitlements, usage limits and billing terms separately.

4. A future increase is announced

The October source capture records Gemini 4 Argon rates doubling from December 1, 2026 and Gemini 3.8 Flash rates doubling from January 1, 2027. These are announcements retained in that edition, not a newly verified forecast. Confirm the current official notice and your applicable terms. An unchanged-workload scenario is arithmetic; actual spending also depends on usage, caching and model choice.