Models
Frontier models.
Production ready.
Access cutting-edge video, image, and text models - carefully selected,rigorously tested, and optimized for real production workflows at scale.
Carefully Tested
Hands-on evaluation before listing
Production Ready
Reliable quality and consistent outputs
Workflow Optimized
Works seamlessly with PomexAI workflows
We keep our catalog focused so you can find the right model faster.
A focused catalog you can trust
MiniMax H3
MiniMax H3 is the MiniMax video route for text-to-video, image-to-video, and first/last-frame workflows, with audio generated natively on every clip. It supports 768P and 2K outputs at a fixed 24 fps, and is billed per second of generated video rather than per token.
Max Duration
DeepSeek V4.1 Flash
DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on output from a 552B-parameter backbone, an asymmetric split that keeps per-token compute low relative to the model's total size. Image understanding is native to the architecture, with visual and text embeddings trained jointly from the start of pre-training rather than added afterward as in the earlier experimental V4 Flash Vision Exp(opens in new tab).
Price
$0.15/$0.6 per M
Context Window
Seedance 2.0
Currently ranked #1 in the Text-to-Video Arena; supports 12 multimodal reference inputs for high consistency.
Price
from $2.4/1M tokens
Max Duration
15s (High FPS)
SkyReels V3
Add a multi-shot generation path for prompts, images, and reference-guided video that requires consistent motion and scene continuity.
Price
$32 input / $32 output
Max Duration
18s
Our evaluation process
Four steps separate a research release from a production-ready model in our catalog.
1. Discover
Track releases from leading research labs, open-source communities, and commercial providers.
2. Benchmark
Test generation quality, inference speed, and API reliability against production requirements.
3. Stress Test
Build real workflows with the model to verify it performs under production constraints at scale.
4. Ship
Only models that consistently deliver production-quality results make it into the catalog.
Can't find the model you need?
Tell us what you're looking for. We're always evaluating new models.