AILIMIT.WATCH

Gemini

Last verified: Jun 6, 2026

1 unannounced

11 sourced changes·64% official·verified against official docs, changelogs & Wayback·last checked Jun 6, 2026

01 / VERDICT

Stable

Low recent constraint volatility — few meaningful restrictions in the last 12 months.

WHY STABLEtrend: stable · drift -1% over 90d · 1 tightening in 90d

Constraint Stability

93/100

higher = fewer recent restrictions

stable-1% / 90d
1 unannounced change

Gemini has stayed stable, with minimal changes to its limits over time.

Most noticeable: Moved to opaque compute-based limits

Recency
65
days since it last tightened
Velocity
2.5
tightenings per year
Silent
50%
of tightenings were unannounced
Reversals
50%
of tightenings later reversed (same dimension)

Gemini constraint metrics — computed from 2 tracked tightenings · median 173 days between tightenings. Original to this site; not published by the vendor.

IMPACT BY PERSONA
Solo / occasional
Brief, occasional use is rarely touched by limit changes.
low
Power user / daily heavy
Exposed to usage-limit, sustained-session, rate / frequency tightening.
medium
Team / shared workflow
Exposed to quota changes across the team.
medium
Startup / cost-scaling
Pricing and usage-based cost have been predictable as volume grows.
low
Enterprise / long-term dependency
Pricing predictability, model continuity and ownership have been stable long-term.
low

02 / ANALYSIS

Gemini's real movement is about transparency — eroding first, then partially recovering under pressure. It isn't a simple tightening: the daily image-generation allowance rose from 30 to 100 in October 2025, and the cheaper AI Plus tier ($7.99/mo) reached the US and 35 markets in January 2026. The limit mechanics, though, moved the wrong way for anyone who values knowing where they stand. In November 2025 the published free allowance of up to five Gemini 3 Pro prompts per day was quietly replaced with vague 'basic access' language and a note that limits 'may change frequently' — a concrete number silently removed. May 2026 formalized the opacity, shifting to compute-based usage that varies by prompt complexity and features, with a five-hour refresh under a weekly cap. Backlash again forced a partial walk-back within days: Flash-Lite was made off-quota and failed requests stopped being charged. A user report captures the effect — a single prompt eating several percent of the five-hour budget, free Pro access disappearing overnight. Recent additions (text-to-video, Meet notes, new Nano Banana models) expand capability without restoring clarity. Gemini's ceiling is real but increasingly unknowable in advance. Plan around the fact that free-tier access to premium models can be cut without a stated figure, and that what counts against your budget is deliberately fuzzy.

03 / CURRENT LIMITS

04 / WHERE IT MATTERS

Which usage patterns the recent changes affect — each line backed by the events driving it.

Sustained sessionsmedium exposure
May 2026Moved to opaque compute-based limits·source
High-frequency usagemedium exposure
May 2026Partial walk-back after complaints·source
May 2026Moved to opaque compute-based limits·source
Nov 2025Free Pro-model limit obscured/cut due to demand·source
Oct 2025Daily image-generation allowance raised 30 → 100·source
Specific-model dependencelow exposure
Nov 2025Free Pro-model limit obscured/cut due to demand·source
Agentic / high-volume costlow exposure
Jun 2026Adds new model options for developers and users.·source

05 / WHAT CHANGED

Technical 4 · Product 4 · Business 2 · Policy 1

  1. May 2026

    Moved to opaque compute-based limits

    technicalImpact High · source

    shipped alongside Adds new text-to-video generation capabilities via Gemini Omni. (Jun 2026) — more for more, not a pure cut.

  2. Nov 2025

    Free Pro-model limit obscured/cut due to demand

    technicalImpact High · unannounced · source
  3. Oct 2025

    Daily image-generation allowance raised 30 → 100

    technicalImpact Low · source · secondary source
▸ View full constraint history (7 events)
  1. Jul 2026

    A newly listed Gemini feature (Gemini Spark) is available in the US only at launch; not a restriction of previously available access.

    policyImpact Low · source · secondary source
  2. May 2026

    Partial walk-back after complaints

    technicalImpact Low · source
  3. May 2026

    Moved to opaque compute-based limits

    technicalImpact High · source

    shipped alongside Adds new text-to-video generation capabilities via Gemini Omni. (Jun 2026) — more for more, not a pure cut.

  4. May 2026

    New $100 Ultra tier; top tier cheaper

    businessImpact Low · source · secondary source
  5. Jan 2026

    Cheaper AI Plus tier reached the US

    businessImpact Low · source
  6. Nov 2025

    Free Pro-model limit obscured/cut due to demand

    technicalImpact High · unannounced · source
  7. Oct 2025

    Daily image-generation allowance raised 30 → 100

    technicalImpact Low · source · secondary source
▸ Other product updates (4) — feature launches, not constraint changes
  1. Jul 2026

    Adds a new organizational feature for study workflows in the Gemini app.

    productImpact Low · source · secondary source
  2. Jun 2026

    Adds new model options for developers and users.

    productImpact Low · source · secondary source
  3. Jun 2026

    Adds new functionality for Google AI Pro and Ultra subscribers in Google Meet.

    productImpact Low · source · secondary source
  4. Jun 2026

    Adds new text-to-video generation capabilities via Gemini Omni.

    productImpact Low · source · secondary source

06 / OUTLOOK

Volatile

Volatile. Last 12 months: 2 pricing, 4 limit/quota, 4 model-availability changes (2 tightenings).

Observed past cadence — not a prediction.

07 / USER SIGNALS

USER SIGNALS — HACKER NEWS · GITHUB · APP STORE

11 complaint mentions · last 180d

11 HN

  • Pricing·327%
  • Performance·327%
  • Limits / quota·218%
  • Other·218%
  • Model removal·19%

Hacker News, GitHub issues & Apple App Store reviews matching complaint keywords — not a scientific sample. See methodology.

08 / ALTERNATIVES

ALTERNATIVES

By constraint history — not capability.

Similar risk:Suno, Replit, ChatGPT, Sora
→ Full comparison matrix

09 / RELATED