Community Announcement - February 27th, 2026

To the Chutes Community,

As state-of-the-art AI models continue to grow in size and hardware requirements, operating them reliably requires significantly more infrastructure than in the past. In order to maintain performance, stability, and long-term sustainability, we are implementing several updates to our public offerings.

These decisions were made carefully and with extensive consideration. They are not taken lightly, and every change is intended to improve overall service quality for our users.

1. Early Access Program Changes

In the early days of Chutes (Subnet 64), we introduced an Early Access perk that provided 200 requests per day for free to participating users. This perk has been active for nearly a year, and we are incredibly grateful to everyone who supported us through it.

However, the program is no longer sustainable in its current form and will be discontinued.

Effective Friday, February 27:

• Access to TEE models will be removed for Early Access users.
• The 200 daily request quota will remain available for all non-TEE models until March 15, at which point the plan will be fully retired.

Transition Options (At No Cost)

Early Access users may visit their account page and choose one of the following:

• One month of a Base subscription, or
• A $5 account credit

We deeply appreciate your early support and want to make this transition as fair as possible.

2. Subscription Usage Model Update

Our current token-agnostic, request-based subscription model is no longer viable given the increasing compute demands of modern frontier models.

To better illustrate the issue, below are the Top 5 highest-usage users in each subscription tier (last 30 days). Displayed is the PAYGO equivalent value received compared to the subscription price paid.

Base — $3/month

Plus — $10/month

Pro — $20/month

What’s Changing

Moving forward and effective immediately, all subscriptions will include a maximum usage allowance equal to 5× the equivalent Pay-As-You-Go value, calculated based on the per-million token pricing of the models used.

This maximum may be enforced across different time intervals — not solely as a single monthly cap.

In other words:

• Limits may apply within shorter rolling time windows (such as per day or per several hours).
• Short-term usage windows may allow a higher multiple of your prorated subscription value for that same time period.
• The overall monthly benefit will remain capped at 5×.

For example, a rolling 4-hour window may permit up to a defined multiple of the equivalent PAYGO value for that 4-hour portion of your subscription, while the total monthly benefit remains capped at 5×.

The specific timing and multipliers of these safeguards may evolve over time. Our goal is to:

Ensure fair access

• Protect platform stability
• Maintain strong value for the vast majority of users

Once the threshold has been reached, usage will transition to standard Pay-As-You-Go pricing.

After analyzing platform usage:

• Approximately 85% of users will see no change
• Roughly 15% of high-usage accounts may notice updated limits

Chutes is often the lowest-cost option of any Pay-As-You-Go provider — and we intend to keep it that way.

3. Base Tier Model Availability Changes

Several high-end frontier models will be removed from the Base subscription tier, including:

• GLM-5
• Kimi K2.5
• Qwen 3.5
• MiniMax M2.5

These models are exceptionally resource-intensive and require continuous infrastructure scaling and optimization. Under the current structure, they are frequently overloaded, resulting in slower performance and reliability issues.

By adjusting access:

• Overall system stability will improve
• Performance will increase
• Paid access users will experience better availability

Without these changes, service quality would continue to degrade — which is not acceptable to us.

Additionally, we are actively searching for additional GPUs for inventory but are currently limited due to a global shortage.

We understand that change can be disruptive, and we sincerely appreciate your understanding. Our goal is to build a sustainable, high-performance inference platform that continues to improve over time.

Thank you for being part of the Chutes community.

Sincerely,

The Chutes Team