ChatGPT vs Gemini
How ChatGPT and Gemini compare on constraint stability — usage limits, pricing, model access and transparency over time.
Constraint stability only — pricing, limits, model availability, transparency. Not a capability comparison.
01 / SIDE-BY-SIDE
| Tool | Verdict | Pricing | Limits | Models | Transparency | 90d | Last change |
|---|---|---|---|---|---|---|---|
| ChatGPT | Stable | Stable100 | Stable99 | Stable93 | Stable100 | 0 | New $100 Pro mid-tier (Apr 2026) |
| Gemini | Stable | Stable100 | Shifting82 | Stable100 | Stable92 | 8 | A newly listed Gemini feature (Gemini Spark) is available in the US only at launch; not a restriction of previously available access. (Jul 2026) |
02 / WHICH CHANGES THE RULES LESS?
Transparency is what separates these two. Gemini has drifted toward opacity: in November 2025 it silently swapped its stated '5 Gemini 3 Pro prompts/day' free allowance for vague 'basic access' language whose limits 'may change frequently,' and in May 2026 it moved to compute-based usage tied to prompt complexity, with a 5-hour refresh up to a weekly cap — walked back only partially days later, making Flash-Lite off-quota and no longer charging for failed requests. Users report 'a single prompt eats several percent of the 5-hour budget' and that 'free Gemini Pro access disappeared overnight.' ChatGPT tightened too, but its limits stay concrete enough to plan around: the GPT-5 launch set Plus at 80 messages per 3 hours in August 2025, backlash doubled that to 160 in September, and Plus still advertises a countable 160-message figure. Its own moves are real — ads on Free and Go in February 2026, legacy models retired that same month — yet the caps remain legible. Both eroded free-tier value under demand and both added cheap tiers in early 2026 (ChatGPT Go at $8 globally, Gemini AI Plus at $7.99 in the US), so entry pricing softened on each side. Gemini's problem is that opaque, compute-metered limits turn daily budgeting into guesswork and its free Pro access proved abruptly removable; ChatGPT's is ads and a narrowing model roster. If you need to know each morning what you can actually do, ChatGPT's numeric caps win — a point about knowing your ceiling, not about which model reasons better.