Legal

Fair Use Policy

Status: 12 August 2026

1. Fair Use Limits

We want to ensure that you can use NENNA.AI intensively without individual users affecting performance for others through extreme traffic spikes. Therefore, certain caps on requests per time window apply to the various model categories.

2. Calculation of Requests (Queries)

In order to account for the different complexity of tasks, we calculate limits based on "standard requests":

1 standard request = An input/output volume of up to 30,000 tokens (approximately 20 pages of PDF).
Most chat interactions correspond to exactly 1 request.
Complex tasks (e.g., analysis of very large PDF documents or extremely long chat histories) are automatically counted as multiple requests.

Example: A document with 110,000 tokens is analyzed. This technically counts as 4 requests (110,000 / 30,000, rounded up) and is deducted from your quota.

3. Categories & Quotas (per user)

Category 1: Standard Models

e.g., Deepseek, Kimi K2.7 – all models not listed under Category 2.Limit: 300 requests every 3 hours.

This means: Practically unlimited working for normal tasks.

Category 2: High-Performance

e.g., Claude Opus 4.7, GPT 5.6 Sol Limit: 100 requests every 3 hours.

This means: Sufficient for intensive deep-work sessions, but prevents abusive automation.

High-Performance Models

all Anthropic OPUS models
Flux 2 Pro
GPT Image 1
Nano Banana Pro
OpenAI GPT 5.5
OpenAI GPT 5.6 Sol

The counters are reset every 3 hours (rolling).