Free Model Quota

1. Overview

MegaNova provides free usage quotas for selected models across supported tiers to enable onboarding, testing, and lightweight production use.

Free quotas are model-specific and tier-scoped, and are designed to:

• Lower initial adoption friction

• Enable developers to validate integrations

• Provide sustainable free access without compromising system stability

This document defines which models include free quota, and how that quota is applied.

2. Free Quota Principles

• Free quota is granted per account, not per request

• Quota is reset on a fixed cycle (daily reset at 00:00 UTC)

• Once free quota is exhausted, usage continues based on the account's billing configuration (if the charge switch is turned on)

• Free quota availability varies by model

3. Supported Models with Free Quota

The following table lists all models that include free quota across different tiers:

Model	Model Size	Type	Free Quota Tier 1	Free Quota Tier 2	Free Quota Tier 3	Free Quota Tier 4
Mistral-Small-3.2-24B-Instruct-2506 (mistralai/Mistral-Small-3.2-24B-Instruct-2506)	24B	Text	50	300	300	1,000
Manta-Mini-1.0 (meganova-ai/manta-mini-1.0)	8B-70B	Text	50	500	500	1,000
Manta-Flash-1.0 (meganova-ai/manta-flash-1.0)	8B-685B	Text	50	500	500	1,000
Manta-Pro-1.0 (meganova-ai/manta-pro-1.0)	8B-1T	Text	0	50	100	200
L3.3-MS-Nevoria-70B (Steelskull/L3.3-MS-Nevoria-70b)	70B	Text	50	300	300	1,000
L3-70B-Euryale-v2.1 (Sao10K/L3-70B-Euryale-v2.1)	70B	Text	50	300	300	1,000
L3-8B-Stheno-v3.2 (Sao10K/L3-8B-Stheno-v3.2)	8B	Text	50	500	500	1,000
Sapphira-L3.3-70B-0.1 (BruhzWater/Sapphira-L3.3-70b-0.1)	70B	Text	50	300	300	1,000
MN-Violet-Lotus-12B (FallenMerick/MN-Violet-Lotus-12B)	12B	Text	50	500	500	1,000
Faster-Whisper-Large-V3 (Systran/faster-whisper-large-v3)	-	Audio	0	5,000	10,000	50,000
Qwen3-Embedding-8B (Qwen/Qwen3-Embedding-8B)	8B	Embedding	50	500	2,000	5,000
BGE-Reranker-V2-M3 (BAAI/bge-reranker-v2-m3)	0.6B	Reranker	50	500	500	1,000
GLM-4.7-Flash (zai-org/GLM-4.7-Flash)	30B MoE	Text	0	50	100	100
Total Free Quota per day			550	4,650	4,800	12,500

Characteristics:

• Optimized for onboarding and prototyping

• Stable latency and predictable cost

• Recommended for experimentation and testing

4. Quota Enforcement Logic

• Free quota is consumed first when a supported model is used

• Quota tracking is model-aware

• Each request reports: Remaining free quota

• When free quota reaches zero, if users turn on the "Allow paid usage after free Quota"

  • Requests continue normally under paid usage

  • No request interruption occurs

5. Tier Interaction

Account Tier	Free Quota Access
Tier 1	Enabled (selected models only)
Tier 2 +	Enabled + higher limits
Team / Pro	Enabled (may vary by contract)
Enterprise	Configurable

Free quota does not affect paid quotas or contracted limits.

6. Best Practices

Use free-quota models for:

• Prototyping

• Chatbots

• Conversational and narrative flows

• Developer testing

Reserve paid models for:

• Complex reasoning

• Long-context conversations

• Multimodal applications

7. Change Management

Model free quota eligibility may evolve based on:

• Infrastructure capacity

• Model performance

• Usage patterns

All updates will be reflected in this document and surfaced through API metadata.