AILIMIT.WATCH

Midjourney

Last verified: Jun 7, 2026

6 sourced changes·50% official·verified against official docs, changelogs & Wayback·last checked Jun 7, 2026

01 / VERDICT

Stable

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

WHY STABLEtrend: loosening · drift +5% over 90d · no recent tightening

Constraint Stability

96/100

higher = fewer recent restrictions

loosening+5% / 90d

Midjourney has eased its constraints recently.

Most noticeable: Video jobs burn ~8x the fast-GPU time of images

Recency
128
days since it last tightened
Velocity
2.7
tightenings per year
Silent
0%
of tightenings were unannounced
Reversals
33%
of tightenings later reversed (same dimension)

Midjourney constraint metrics — computed from 3 tracked tightenings · median 140 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
Daily-limit, session and rate constraints have been steady.
low
Team / shared workflow
Exposed to usage-based cost changes across the team.
medium
Startup / cost-scaling
Exposed to usage-based cost as usage scales.
medium
Enterprise / long-term dependency
Pricing predictability, model continuity and ownership have been stable long-term.
low

02 / ANALYSIS

Midjourney meters by GPU time, and the friction is that every new capability layered onto that meter costs disproportionately in fast-hours. The V1 video model (June 2025) made a video job run about 8x a typical image job — enough, the user report confirms, to drain a Basic plan quickly. Flagship launches have repeatedly arrived gated and degraded: V8 Alpha (March 2026) was confined to alpha.midjourney.com with Relax mode — the unmetered valve — unsupported, and sref/Moodboard jobs costing and running 4x, so 'unlimited' renders effectively didn't apply for that stretch. V8.1 later reached midjourney.com with native 2K HD, but HD burns ~33% more GPU time than SD; it became the default in June 2026, with a draft mode added soon after. Sitting over all of it is the June 2025 Disney/NBCUniversal/DreamWorks copyright suit, which seeks injunctive relief that could force output-restricting changes to the image service. Capability keeps shipping — but each new tier of it (video, HD, alpha models) is metered harder or temporarily stripped of the unmetered Relax path lighter plans lean on. Judged strictly on constraint stability rather than image quality: headline 'unlimited Relax' claims hold mainly for mature models on the main site; budget video and new models at multiples of image cost, expect the newest model to route through an alpha site without Relax for weeks, and note the unresolved litigation over what outputs stay permitted.

03 / CURRENT LIMITS

04 / WHERE IT MATTERS

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

Agentic / high-volume costmedium exposure
Mar 2026New model alpha-gated; Relax (unmetered) unavailable, styling jobs cost 4x·source
Jun 2025Video jobs burn ~8x the fast-GPU time of images·source
Sustained sessionslow exposure
Mar 2026New model alpha-gated; Relax (unmetered) unavailable, styling jobs cost 4x·source
Specific-model dependencelow exposure
Jun 2026Users are transitioned to the V8.1 model as the new default standard.·source
Mar 2026New model alpha-gated; Relax (unmetered) unavailable, styling jobs cost 4x·source

05 / WHAT CHANGED

Technical 0 · Product 4 · Business 1 · Policy 1

  1. Mar 2026

    New model alpha-gated; Relax (unmetered) unavailable, styling jobs cost 4x

    productImpact Low · source
  2. Jun 2025

    Video jobs burn ~8x the fast-GPU time of images

    businessImpact Medium · source
  3. Jun 2025

    Copyright suit; possible forced content/output restrictions

    policyImpact Medium · source
▸ View full constraint history (5 events)
  1. Jun 2026

    Users are transitioned to the V8.1 model as the new default standard.

    productImpact Low · source · secondary source
  2. Apr 2026

    V8.1 generally available; HD ~33% more GPU time than SD

    productImpact Low · source · secondary source
  3. Mar 2026

    New model alpha-gated; Relax (unmetered) unavailable, styling jobs cost 4x

    productImpact Low · source
  4. Jun 2025

    Video jobs burn ~8x the fast-GPU time of images

    businessImpact Medium · source
  5. Jun 2025

    Copyright suit; possible forced content/output restrictions

    policyImpact Medium · source
▸ Other product updates (1) — feature launches, not constraint changes
  1. Jun 2026

    Adds new functionality for V8.1, expanding user capabilities.

    productImpact Low · source · secondary source

06 / OUTLOOK

Active

Active. Last 12 months: 0 pricing, 0 limit/quota, 4 model-availability changes (1 tightening).

Observed past cadence — not a prediction.

07 / USER SIGNALS

USER SIGNALS — HACKER NEWS · GITHUB · APP STORE

17 complaint mentions · last 180d

17 HN

  • Pricing·1059%
  • Performance·318%
  • Model removal·212%
  • Limits / quota·16%
  • Other·16%

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