Seedance 2.0 Mini video from $0.008/sec.

Built for AI comic series and short videos, with cost-effective generation at as low as 40% of official rates.

Crun

English
Login

Text to VideoImage to VideoReference to Video

Seedance 2.5

The maximum generation time in a single session can be up to 30 seconds, while supporting up to 50 reference materials, including up to 30 images, 10 videos, and 10 audio clips. This expanded multi-reference capability helps maintain consistent character, visual style, and scene continuity in long AI videos.

Text to ImageImage to Image

Nano Banana 2.1

Nano Banana 2.1 delivers significantly improved image quality and detail while maintaining ultra-fast generation speeds. It offers powerful prompt understanding, precise multilingual text layout and rendering, realistic lighting and textures, and excellent subject consistency. From rapid concept testing to high-quality commercial visual content creation in up to 4K resolution, it handles it all with ease.

Text to ImageImage to Image

GPT Image 2.5

Delivers 50% lower latency, pixel-level targeted multi-turn editing, rock-solid subject fidelity, and native transparent PNG output. Integrated with Flare and Sunburst models to empower fast, high-precision visual creation.

Text to VideoImage to VideoReference to Video

Google Veo 3.1

Google Veo 3.1 is a top-tier video model that integrates high-fidelity 1080p video quality, native audio-visual synchronization, and deep director-level control, achieving a cinematic, long-duration AI narrative experience through outstanding subject consistency.

Text to VideoImage to VideoReference to Video

Wan 3.0

Wan 3.0 Video is Alibaba's flagship AI video foundation model. Supporting Text-to-Video, Image-to-Video, and Reference-to-Video (R2V), Wan 3.0 delivers 1080p cinematic output, realistic physical world simulation, smooth camera control, and exceptional character consistency.

Text to VideoImage to VideoReference to Video

Vidu Q3

Vidu Q3 is the world’s first video generation model supporting 16-second synchronized audio and video. It enables seamless one-shot creation in 1080P, with precise camera control and multilingual output—built for cinematic storytelling.

Text to ImageImage to Image

Seedream 5.0 Pro

Built for professional design workflows with advanced design understanding, accurate multilingual text rendering, precise local editing, and high-quality image generation for posters, marketing visuals, illustrations, and commercial creative assets.

All Models

Name: AI API Models
Brand: Crun

94 relevant models were found.

Type:

Providers:

AIsphere

Pixverse V6 Max

$0.01485-45%

4 Credits / s

Text to VideoImage to Video

AIsphere

Pixverse V6 Max

Create videos at up to 1080p with native audio, multi-clip storytelling, flexible aspect ratios, and consistent visuals guided by your prompts and references.

Kling

Kling 4.0

17 Credits / s

Text to VideoImage to VideoFrames to VideoReference to Video

Kling

Kling 4.0

Multimodal video generation with text-to-video, image-to-video, keyframe interpolation, and Omni reference, giving control over character action, camera movement, audio, and lip sync.

Black Forest Labs

FLUX 3 Klein

1.45 Credits / image

Text to ImageImage to Image

Black Forest Labs

FLUX 3 Klein

A lightweight 5.6B multimodal image model delivering ultra-fast 1-16 step generation, multi-reference consistency for up to 10 images, and precise instruction-based editing.

OpenAI

GPT-6.1 Sol

398.72 Credits / 1m input tokens

1993.6 Credits / 1m output tokens

LLM

OpenAI

GPT-6.1 Sol

Near-flagship agentic model: coding on par with GPT-6 Astra (DeepSWE v1.1), computer use within 2 points (OSWorld 2.0), 1.05M-token context and five reasoning-effort levels.

OpenAI

GPT-6 Luna

19.936 Credits / 1m input tokens

99.68 Credits / 1m output tokens

LLM

OpenAI

GPT-6 Luna

GPT-6 Luna is OpenAI's fastest and lightest high-efficiency reasoning model: 1.05M-token context, reasoning effort from none to max, 66.6% on DeepSWE v1.1, 52.7% on OSWorld 2.0 and ~157 tokens/s for high-volume agentic work.

Anthropic

Claude Sonnet 5.5

403.2 Credits / 1m input tokens

2016 Credits / 1m output tokens

LLM

Anthropic

Claude Sonnet 5.5

Runs 30%+ faster than Claude Sonnet 5, delivering Opus-level agentic coding and knowledge work. As the fastest Sonnet model yet, it also brings a sharp eye for design and ships with cyber safeguards from day one.

Anthropic

Claude Opus 5.5

806.4 Credits / 1m input tokens

4032 Credits / 1m output tokens

LLM

Anthropic

Claude Opus 5.5

Performs at the level of Claude Fable 5.1 on most tasks, and with stronger tool use, 30%+ faster output, and Anthropic's most secure coding agent, it is built for large-scale refactors, financial analysis, and production AI agents.

Google

Gemini 4 Argon

313.6 Credits / 1m input tokens

1568 Credits / 1m output tokens

LLM

Google

Gemini 4 Argon

Google's flagship reasoning model with a 1M token context and 64k output, engineered for long-horizon agents, complex software engineering, and deep multimodal reasoning.

Vidu

Vidu Q4

$0.028125-38%

4 Credits / s

Image to VideoReference to Video

Vidu

Vidu Q4

Supports Image-to-Video and advanced Reference-to-Video with up to 15 image references, featuring native audio sync, precise lip-sync, and consistent character dynamics in up to 4K.

Anthropic

Claude Haiku 5.5

20.16 Credits / 1m input tokens

100.8 Credits / 1m output tokens

LLM

Anthropic

Claude Haiku 5.5

Claude Haiku 5.5 delivers exceptional speed and cost efficiency with a 1M token context, 128k output, and adaptive thinking, engineered for high-volume agent workflows and rapid coding.

VAST

Tripo H3.1

$0.0938-41%

21 Credits / 3d

VAST

Tripo H3.1

Tripo H3.1 quickly generates detailed, high-quality, production-ready 3D models for games, rendering, and 3D printing through high-density geometry and PBR material support.

VAST

Tripo P1.0

$0.0938-41%

21 Credits / 3d

VAST

Tripo P1.0

Tripo P1.0 generates lightweight, low-poly 3D meshes with clean topology in about 2 seconds. Designed for production workflows such as game development, real-time rendering, and XR, it reduces the need for manual topology optimization.