Skip to main content

Agnes AI Access Plans, API Keys, RPM Limits, and Subscription Quotas FAQ

Effective date: 2026-06-22 Version: Current public reference version This document explains the current Agnes AI access plans, API key types, RPM limits, and subscription quotas. The values below are current public reference values as of 2026-06-22. They are operational limits, not permanent guarantees. Agnes AI may adjust these limits based on infrastructure capacity, service stability, product updates, abuse prevention, or commercial policy changes.

1. Overview

Agnes AI currently supports three main access types:
Access typeDescription
Free / default usersUsers without a Token Plan subscription or Enterprise certification.
Enterprise-certified usersUsers who have completed Enterprise certification and receive higher baseline access limits.
Token Plan usersUsers who subscribe to a Token Plan and receive significantly higher model access limits and subscription quotas.
A single user may have more than one access type at the same time. For example, a user can:
  • Use Free / default keys.
  • Complete Enterprise certification and use Enterprise-certified keys.
  • Subscribe to a Token Plan and use Token Plan keys.
  • Be both a Token Plan user and an Enterprise-certified user.
Each access type uses its own API key type and its own corresponding limit pool.

2. Current Access Limits

2.1 Free / Default Users

Free / default users currently have the following access limits:
Model typeCurrent limit
Text models20 actual RPM
Image modelsResolution-specific RPM limits apply
Video models20 actual RPM
Free access is intended for basic usage, evaluation, lightweight testing, and low-frequency integrations.

2.2 Enterprise-Certified Users

Enterprise-certified users currently have higher baseline access limits:
Model typeCurrent limit
Text models40 actual RPM
Image modelsHigher resolution-specific RPM limits apply
Video models40 actual RPM
Enterprise certification improves the user’s baseline access level. Enterprise certification does not automatically include Token Plan subscription quotas unless the user also subscribes to a Token Plan.

2.3 Token Plan Users

Token Plan users currently have significantly higher access limits:
Model typeCurrent limit
Text models1,000 actual RPM
Image modelsHigher 1K and 2K image RPM limits apply
Video models100 actual RPM
Token Plan is recommended for users who need higher concurrency, more stable production usage, or frequent model calls.

3. Token Plan Subscription Quotas

Token Plan users also receive subscription quotas. These quotas are separate from RPM limits.
Planagnes-2.0-flashagnes-image-2.0/2.1-flashagnes-video-v2.0
Starter1,500 requests per 5 hours; 15,000 requests per week4,000 images per day500 seconds per day
Plus7,500 requests per 5 hours; 75,000 requests per week4,000 images per day500 seconds per day
Pro30,000 requests per 5 hours; 300,000 requests per week4,000 images per day500 seconds per day

4. RPM Limits vs Subscription Quotas

RPM limits and subscription quotas are different.

RPM Limits

RPM means requests per minute. RPM limits control how many requests can be processed within one minute. For example:
  • Free / default users have 20 actual RPM for text models.
  • Enterprise-certified users have 40 actual RPM for text models.
  • Token Plan users have 1,000 actual RPM for text models.

Subscription Quotas

Subscription quotas control how much usage is available within a specific time window. For example:
  • Starter users can use 1,500 agnes-2.0-flash requests per 5 hours and 15,000 requests per week.
  • Plus users can use 7,500 requests per 5 hours and 75,000 requests per week.
  • Pro users can use 30,000 requests per 5 hours and 300,000 requests per week.

Both Limits Apply

A user must stay within both:
  1. The RPM limit.
  2. The subscription quota.
For example, a Token Plan Pro user may have 1,000 actual RPM for text models, but the user is still limited by the Pro plan quota of 30,000 requests per 5 hours and 300,000 requests per week.

5. API Key Types

Each access type has its own API key type. Current API key types include:
API key typeWho can use itLimit behavior
Free / default keyAll usersShares the Free / default RPM pool
Enterprise-certified keyEnterprise-certified usersShares the Enterprise-certified RPM pool
Token Plan keyToken Plan subscribersShares the Token Plan RPM and subscription quota pool
Keys from different access types can be used independently.

6. API Key Limit Pooling

Limits are applied by key type, not by each individual key. Creating multiple keys of the same type does not increase the total available RPM or quota.

Example 1: Multiple Free Keys

If a Free user creates two Free keys, both keys share the same Free / default RPM pool. The user does not receive a separate 20 RPM limit for each key.
KeyKey typeLimit pool
Key AFree / default keyFree / default RPM pool
Key BFree / default keyFree / default RPM pool
Both keys share the same Free / default RPM limit.

Example 2: Multiple Token Plan Keys

If a Token Plan user creates multiple Token Plan keys, those keys share the same Token Plan RPM and subscription quota pool.
KeyKey typeLimit pool
Key AToken Plan keyToken Plan RPM and subscription quota pool
Key BToken Plan keyToken Plan RPM and subscription quota pool
Creating more Token Plan keys does not increase the total Token Plan quota.

Example 3: User With Free, Enterprise, and Token Plan Keys

A user may have:
  • 2 Free / default keys.
  • 1 Enterprise-certified key.
  • 2 Token Plan keys.
The limits are applied as follows:
Key groupHow limits apply
2 Free / default keysShare one Free / default RPM pool
1 Enterprise-certified keyUses the Enterprise-certified RPM pool
2 Token Plan keysShare one Token Plan RPM and subscription quota pool
Each key type has its own corresponding limit pool. Keys of the same type share limits. Keys of different types use different limit pools.

7. Can One User Have Multiple Access Types?

Yes. A user can have multiple access types at the same time. For example:
User statusAvailable key types
Free userFree / default keys
Enterprise-certified userFree / default keys and Enterprise-certified keys
Token Plan userFree / default keys and Token Plan keys
Token Plan + Enterprise-certified userFree / default keys, Enterprise-certified keys, and Token Plan keys
This means that a Token Plan user may also complete Enterprise certification. In that case, the user may use Token Plan keys and Enterprise-certified keys independently. The user may also continue to use Free / default keys.

8. Are Limits Shared Across Different Key Types?

No. Limits are shared only within the same key type. For example, if a user has both Free keys and Token Plan keys:
Key typeLimit pool
Free / default keysShare the Free / default RPM pool
Token Plan keysShare the Token Plan RPM and subscription quota pool
The Free key does not consume the Token Plan key’s RPM pool. The Token Plan key does not consume the Free key’s RPM pool. However, all keys of the same type share the same corresponding limit pool.

9. What Does “Actual RPM” Mean?

“Actual RPM” refers to the effective number of requests that can be processed per minute. Some systems may use higher request admission thresholds for traffic control, buffering, or abuse prevention. However, the values listed in this document represent the effective public RPM reference limits.

10. Image Model Limits

Image model limits are resolution-specific. Higher resolutions may have lower RPM limits because they consume more infrastructure resources. For example:
  • 1K image generation may support higher RPM.
  • 2K image generation may support lower RPM than 1K.
  • 3K and 4K image generation may have stricter RPM limits.
Token Plan users receive higher 1K and 2K image RPM limits compared with Free / default users. For detailed per-resolution image model RPM limits, please refer to MODEL_CATALOG.md.

11. Video Model Limits

Video models may be limited by both request frequency and generated video duration. For Token Plan users, the current agnes-video-v2.0 quota is:
PlanVideo quota
Starter500 seconds per day
Plus500 seconds per day
Pro500 seconds per day
In addition, video model RPM limits still apply. Current video RPM limits are:
User typeVideo model RPM
Free / default users20 actual RPM
Enterprise-certified users40 actual RPM
Token Plan users100 actual RPM

12. What Happens If a User Exceeds the RPM Limit?

If a user exceeds the RPM limit, the request may be rejected or rate-limited. The user may receive an error such as:
text
429 Too Many Requests
To resolve this, the user can:
  • Reduce request frequency.
  • Add retry and backoff logic.
  • Use a higher access tier if eligible.
  • Subscribe to a Token Plan for higher RPM limits.
  • Complete Enterprise certification for higher baseline limits.

13. What Happens If a Token Plan User Exceeds the Subscription Quota?

If a Token Plan user exceeds the subscription quota, additional requests may be rejected until the quota resets. For example:
  • If the 5-hour text request quota is exhausted, the user must wait for the next 5-hour window.
  • If the weekly text request quota is exhausted, the user must wait until the weekly quota resets.
  • If the daily image or video quota is exhausted, the user must wait until the daily quota resets.
RPM limits and subscription quotas are checked separately.

14. Does Creating More API Keys Increase the Limit?

No. Creating more API keys under the same key type does not increase the total available RPM or quota. For example:
  • 1 Free key and 5 Free keys share the same Free / default RPM pool.
  • 1 Token Plan key and 5 Token Plan keys share the same Token Plan RPM and subscription quota pool.
  • 1 Enterprise-certified key and 5 Enterprise-certified keys share the same Enterprise-certified RPM pool.
API keys are intended for access control, application separation, and operational management. They are not intended to multiply usage limits.

Free / Default Users

Free / default access is suitable for:
  • Basic testing.
  • Lightweight personal usage.
  • Initial API exploration.
  • Low-frequency integrations.

Enterprise-Certified Users

Enterprise-certified access is suitable for:
  • Business users.
  • Internal testing.
  • Higher baseline usage.
  • Teams that need higher limits than Free / default users.

Token Plan Users

Token Plan access is suitable for:
  • Production usage.
  • Higher-frequency API calls.
  • Agent workflows.
  • Multi-user applications.
  • Applications that require higher throughput.
  • Applications that use image or video generation frequently.

16. Best Practices

To avoid rate limit or quota issues, developers should:
  • Use the correct API key type for the workload.
  • Avoid using Free / default keys for production workloads.
  • Add retry and exponential backoff for 429 errors.
  • Monitor request frequency.
  • Monitor daily, 5-hour, and weekly usage.
  • Separate environments with different keys, such as development, staging, and production.
  • Avoid creating multiple keys for the purpose of bypassing limits.
  • Use Token Plan keys for high-frequency or production traffic.
  • Use Enterprise-certified keys when higher baseline access is required.

17. FAQ

Q1: Can a Free user create multiple API keys?

Yes. A Free user can create multiple Free / default keys. However, all Free / default keys share the same Free / default RPM pool. Creating more Free keys does not increase the user’s RPM limit.

Q2: Can a Token Plan user still use Free keys?

Yes. A Token Plan user may still have and use Free / default keys. Free / default keys use the Free / default RPM pool. Token Plan keys use the Token Plan RPM and subscription quota pool.

Q3: Can a Token Plan user also become Enterprise-certified?

Yes. A Token Plan user can also complete Enterprise certification. In this case, the user may have access to:
  • Free / default keys.
  • Enterprise-certified keys.
  • Token Plan keys.
Each key type can be used independently and follows its own limit pool.

Q4: Do Enterprise-certified users automatically get Token Plan quotas?

No. Enterprise certification and Token Plan subscription are separate. Enterprise certification provides higher baseline RPM limits. Token Plan provides higher RPM limits and subscription quotas. A user must subscribe to a Token Plan to receive Token Plan quotas.

Q5: Are Free key limits and Token Plan key limits shared?

No. Free keys and Token Plan keys use separate limit pools. A Free key does not consume the Token Plan quota. A Token Plan key does not consume the Free / default RPM pool.

Q6: Are multiple Token Plan keys counted separately?

No. Multiple Token Plan keys under the same user share the same Token Plan RPM and subscription quota pool. Creating additional Token Plan keys does not increase the total available quota.

Q7: Are image quotas counted by request or by generated image?

For Token Plan users, image quota is currently counted by generated images. The current quota for agnes-image-2.0/2.1-flash is 4,000 images per day. Resolution-specific RPM limits may also apply.

Q8: Are video quotas counted by request or by duration?

For Token Plan users, video quota is currently counted by generated video duration. The current quota for agnes-video-v2.0 is 500 seconds per day. Video model RPM limits may also apply.

Q9: Can limits change in the future?

Yes. The limits in this document are current public reference values effective from 2026-06-22. Agnes AI may update these limits based on infrastructure capacity, model availability, abuse prevention, service stability, or product policy changes.

18. Summary

Agnes AI currently supports Free / default users, Enterprise-certified users, and Token Plan users. Each access type uses a separate API key type. Multiple keys of the same type share the same limit pool. Different key types use different limit pools. Token Plan users can also complete Enterprise certification and may continue to use Free / default keys. RPM limits control request frequency. Subscription quotas control total usage within a specific time window. All current values in this document are effective as of 2026-06-22 and may be updated in the future.