Helios-Base

A video generation model capable of creating video from text (T2V), images (I2V), and video (V2V), designed for real-time use and long-duration applications. It can generate video sequences lasting several minutes at a frame rate of 19.5 frames per second (FPS) using a single H100 GPU. The uniqueness of the model lies in its avoidance of traditional anti-drift methods (e.g., self-forcing, error-banks) and standard acceleration techniques (KV-cache, causal masking), all while maintaining video quality and synchronization.

Key Features:

Requirements: 80 GB VRAM across 4 GPUs (fp32).
Group Offloading: Reduces VRAM consumption to ~6 GB, enabling execution on GPUs with limited memory.
Multi-GPU Parallelism: Support for Ulysses Attention, Ring Attention, and other mechanisms for distributed computation.
Inference Pipelines: Integration with Diffusers, vLLM-Omni, SGLang-Diffusion for video generation through modular pipelines.
Settings: The number of frames must be a multiple of 33 (e.g., 240 frames = 60 seconds at 24 FPS).

---

The model is a component of the video generation pipeline, consisting of:

• Text encoder: ~5.7B parameters,
• Transformer: ~28.7B parameters,
• Transformer_init: ~14B parameters,
• VAE: ~127M parameters, 

Total: ~48.7B parameters

Announce Date: 23.02.2026
Parameters: 14B
Developer: BestWishYsh
Diffusers Version: 0.37.0
vLLM-Omni Version: 0.18.0
License: Apache 2.0

Use our pre-built public endpoints for free to test inference and explore Helios-Base capabilities. You can obtain an API access token on the token management page after registration and verification.

Rent your own physically dedicated instance with hourly or long-term monthly billing.

We recommend deploying private instances in the following scenarios:

• maximize endpoint performance,
• enable full context for long sequences,
• ensure top-tier security for data processing in an isolated, dedicated environment,
• use custom weights, such as fine-tuned models or LoRA adapters.

Contact our dedicated neural networks support team at nn@immers.cloud or send your request to the sales department at sale@immers.cloud.