AI Chat · xAI
✓ Verified Jul 2026
Last audited: 2026-07-05
Grok:
What the pricing page doesn't tell you
Based on xAI support emails, community reports, piunikaweb.com reporting, and third-party benchmarks — not legal advice or official statements.
Hidden Behaviors
documented & verified
Transparency
22%
of key policies disclosed upfront
Last Verified
Jul 2026
$30/mo · SuperGrok
Hidden Behaviors
6 documented
Behaviors confirmed through xAI support emails, community reports, and investigative coverage — not disclosed on the SuperGrok pricing page.
Video caps revised downward silently — support emails contradict each other
▾
Official
No official announcement of cap changesxAI made no public blog post or pricing page update when video generation limits changed. The only evidence comes from user-shared support emails with conflicting figures.
Observed (community)
April email: "100+ videos per window" → May email: "80 per 12h"A support email in April stated Heavy users could generate "over 100" videos per reset window. A May email revised this to more than 80 per 12 hours — a reduction with no public notice. Standard SuperGrok dropped to more than 20 videos per 24 hours.
Estimated (not verified)
Effective cap far lower at 720pUsers generating at 720p resolution report hitting limits after 10–15 videos — far below theoretical maximums. Quality stepdown to 480p occurs after the soft limit is reached.
Failed and moderated generations count against your daily cap
▾
Official
Confirmed in support email — not on pricing pagexAI confirmed via support email that moderated or failed image and video generations still count toward the daily cap. This is not disclosed anywhere on the SuperGrok pricing or features page.
Observed (community)
Users lose quota on innocent prompts that get flaggedGrok Imagine's moderation system flags even benign prompts. Each flagged attempt permanently reduces the daily allowance. Users report burning through quotas faster than expected due to moderation failures.
Estimated (not verified)
~10–20% of effective quota lost to moderationCommunity estimates suggest aggressive moderation settings in 2026 mean power users lose a significant fraction of their daily image/video cap to moderated attempts.
Fair use policy is intentionally vague — throttling triggered by usage patterns, not just volume
▾
Official
Fair use defined only as "reasonable and not abusive"xAI's fair use policy uses deliberately vague language. No thresholds are published. xAI reserves the right to throttle or suspend accounts that exceed "reasonable" patterns — even if under published caps.
Observed (community)
Per-session burst limits trigger before daily cap is reachedSending 10 rapid prompts in under 60 seconds triggers a temporary per-session cap — even with daily allowance remaining. Response times slow from ~2s to 8–10s around 180 prompts/hour on SuperGrok.
Estimated (not verified)
Throttling 40% more frequent during peak hoursCommunity reports suggest Grok throttles more aggressively between 2–10 PM UTC. Power users recommend scheduling heavy tasks outside this window to reduce effective throttling by up to 40%.
Heavy tier upgrade upsell triggered precisely when SuperGrok limits hit — community calls it predatory
▾
Official
$99 discounted upgrade pushed at limit triggerxAI displays a $99 discounted upgrade offer to the Heavy tier precisely when SuperGrok rate limits are triggered. The timing is intentional by design.
Observed (community)
Highly upvoted community criticismMultiple highly-upvoted community posts document the pattern and describe it as an aggressive upsell tactic. Users report seeing the upgrade prompt specifically while their work is blocked.
Estimated (not verified)
SuperGrok Heavy: ~$300/mo (not clearly advertised)The SuperGrok Heavy tier at full price represents a significant jump from the $30/mo SuperGrok. Pricing for Heavy is not prominently displayed before users hit the limit trigger.
Image + video share one combined daily pool — not separate quotas
▾
Official
Not disclosed separately on pricing pagexAI does not clearly explain that image and video generation share the same daily operation pool. Users comparing plans may assume they receive separate quotas for each media type.
Observed (community)
Video spikes drain image capacity immediatelyVideo generation consumes compute at a dramatically higher rate than static images. Generating a handful of 720p videos can exhaust the daily image allocation before noon.
Estimated (not verified)
Standard SuperGrok: ~200 image equivalents/day combinedAt typical rates, the shared pool supports roughly 200 image-equivalent operations per day on standard SuperGrok — but one video clip can consume 10–20 image equivalents.
Voice mode limit is per-minute and undisclosed — resets on billing date, not daily
▾
Official
Voice mode limits not published on pricing pageSuperGrok includes Voice Mode but the pricing page does not quantify the monthly voice allowance or explain that it resets on the billing date rather than daily.
Observed (community)
~90 minutes/month on SuperGrok; sessions max 10 minCommunity testing estimates ~90 minutes of voice per month on standard SuperGrok, with sessions auto-ending after 10 minutes. Users relying on Grok for long drives or night shifts find the cap renders voice mode impractical.
Estimated (not verified)
Voice burns 2,000–3,000 tokens/minute internallyEach minute of voice mode consumes backend tokens that contribute to the platform's throttling calculations, even though voice minutes and text prompts use separate visible quotas.
Real Limits — SuperGrok Standard
⚠ Moving Target Alert
Limits change without announcement. April vs May 2026 support emails show different caps for the same plan. No official changelog exists for limit adjustments.
Text Prompts
~1,000/day soft cap — after which throttled to free-tier speeds (not blocked)
Image Gen
~50–60/day (shared pool with video) · failed generations count
Video Gen
~10–20/day at standard quality · drops to 480p after soft limit
Voice Mode
~90 min/month · resets on billing date · sessions auto-end at 10 min
DeepSearch
~50 queries/day · Think Mode capped at ~30 sessions/day
Reset Cycle
24-hour rolling window for most limits — not fixed midnight reset
Burst Limit
Per-session cap triggers after ~10 rapid prompts in 60 seconds — even if daily allowance unused
Transparency Score
Usage Limits
No fixed limits published. Caps change via support email with no public announcement. Failed generation count-against-quota is confirmed only in support correspondence.
Source: piunikaweb.com + xAI support emails
Privacy & Data
Data retention and training use practices are partially disclosed. The relationship between X.com platform data and Grok training is not clearly explained to subscribers.
Source: xAI privacy policy + community analysis
Throttling & Speed
No throttling documentation. Gradual speed reduction instead of hard errors makes limits invisible until users benchmark their own sessions. Peak-hour throttling is undisclosed.
Source: jingrey.com testing + community reports
Billing Clarity
Plan pricing is disclosed. However, the upgrade upsell displayed exactly at limit-hit is an aggressive tactic. Heavy tier pricing is not prominently displayed before users need it.
Source: piunikaweb.com + community reports
Policy Stability
Limits changed between April and May 2026 with no announcement. xAI employee confirmed limits can "fluctuate daily." No changelog or advance-notice commitment exists.
Source: piunikaweb.com May 2026 reporting
Overall Transparency Score
Lowest score in this database. No published limits, no changelog, no advance notice on changes.
22%
Other Services
Grok (xAI)
viewing now
ChatGPT
view report →
Gemini
view report →
Claude
view report →
Perplexity
view report →
Frequently Asked
Does Grok SuperGrok have published rate limits?
No. xAI does not publish fixed rate limits. Limits are described as dynamic and subject to change based on infrastructure. What we know comes from support emails and community testing — not official documentation.
Do failed Grok generations count against your quota?
Yes. xAI confirmed via support email that moderated or failed image and video generations still count against your daily cap — meaning you lose quota even when the system rejects your prompt.
How do Grok's limits compare to ChatGPT and Claude?
SuperGrok offers more daily text messages than ChatGPT Plus (~1,000 vs ~40/3 hours), but its image and video generation limits are stricter, its voice mode is far more restricted, and its transparency score is the lowest of any service in this database.
What triggers Grok's per-session burst limit?
Sending roughly 10 or more rapid-fire prompts in under 60 seconds triggers a temporary per-session slowdown — even if you haven't used your daily allowance. This burst limit is separate from the daily cap and is not documented anywhere officially.
Get alerted when policies change
We track silent limit cuts before most users notice them.