Kling 3

happyhorse-1.1

HappyHorse 1.0 (Happy Horse) is Alibaba's open-source video generation model from the ATH/Taotian division's Future Life Lab - and its standout capability is native multi-shot narrative: t...

happyhorse-1.1

MiniMax

MiniMax H3 Max

MiniMax H3 Max is the maximum-fidelity configuration of MiniMax's H3 video family - the full three-stage pipeline that refines your instructions, generates the audio-video clip, and rebuilds it at native 2K instead of relying on an upscale.

MiniMax H3 Max

Muse Spark 1.3 Contributor

The Muse Spark 1.3 Contributor tier is not a different model - it is a different deal. You get Meta's Muse Spark 1.3 (1M-token context, agent-grade coding, DeepSWE v1.1 at 75.4%) at $0.10 per million input tokens and $0.20 per million output tokens, roughly one-twelfth the standard rate. The price of the discount is your data: prompts and code outputs may be used to train Meta's models.

Muse Spark 1.3 Contributor

Tencent

Tencent Hy4 Preview

Hy4 preview is Tencent Hunyuan's new-generation flagship - a 770-billion-parameter MoE with 49B active parameters and a context window beyond 1M tokens - released and open-sourced on August 28, 2026. Tencent calls it a model "born for productivity," and the internal evidence backs the framing: in a blind test with 163 Tencent domain experts evaluating 203 real engineering tasks, Hy4 preview scored 2.99/4 - ahead of both GLM 5.3 (2.92) and Kimi K3 (2.94).

Tencent Hy4 Preview

Deepseek

DeepSeek V4.1 Flash

DeepSeek V4.1 Flash is the smallest - and architecturally the most unusual - model in DeepSeek's new V4.1 series. Built on a 552B-parameter MoE with an asymmetric Causal-Encoder-Decoder design (roughly 8B active on input, 16B on output), it treats reading and generating as different jobs with different compute budgets - and adds native multimodal vision, KV-cache compression to the point of transforming agent economics, and a benchmark profile that reportedly surpasses the previous generation's Pro model.

DeepSeek V4.1 Flash

Muse Spark 1.3

Muse Spark 1.3 is Meta's latest large language model, positioned as the company's breakthrough in coding and agentic work - and priced to make developers notice. It pairs a 1M-token context window with long-horizon agent workflows that collect their own context, notice gaps in their plans, and adjust without human babysitting.

Muse Spark 1.3

Alibaba

Qwen Image 3.0 Pro

Qwen-Image 3.0 Pro is Alibaba Tongyi's third-generation image foundation model, and it attacks image generation from an unusual direction: layout and text precision rather than artistic flourish. It accepts prompts up to 4.5K tokens, renders 10-pixel small text legibly, handles 12 languages natively, and generates complex structured visuals - nine-grid infographics, nested UI mockups, academic figures - in a single pass.

Qwen Image 3.0 Pro

XAI

Grok Imagine Video 1.5

Grok Imagine Video 1.5 is SpaceXAI's image-to-video model, built on the company's Aurora autoregressive generation engine. It animates a single still image - or a text prompt - into short video with natively synchronized audio, ranks #1 on the Arena.ai image-to-video leaderboard, and generates a 6-second 720p clip in roughly 25 seconds in Fast mode.

Grok Imagine Video 1.5

XAI

Grok Imagine Image 2.0

Grok Imagine Image 2.0 is SpaceXAI's second-generation image model, and its design philosophy is stated in the first sentence of the release: it treats editing as a first-class capability, not a fallback. Where most models invite you to re-roll the dice, Imagine 2.0 hands you a magic wand, a segmentation tool, and a background remover - and keeps your subject consistent through multiple revision rounds.

Grok Imagine Image 2.0

Model

Gemini 3.8 Flash

Gemini 3.8 Flash is Google DeepMind's newest reasoning and coding model - and this release branches the Flash line for the first time. Alongside the standard 3.8 Flash, Google introduced 3.8 Flash Cyber, a dedicated variant for cybersecurity with top-tier vulnerability discovery and automated patch repair, available only to certified defenders through the Fairwind Program.

Gemini 3.8 Flash

ByteDance

Seedance 2.0 Mini

Seedance 2.0 Mini is the compact tier of ByteDance's Seedance 2.0 family - a video generation model built on the same unified multimodal audio-video joint generation architecture as its larger siblings, tuned for speed and cost. It accepts text, images, audio, and video inputs, references up to 12 files per generation, and delivers native synchronized audio with the picture.

Seedance 2.0 Mini

OpenAI

GPT image 2.5 Flare

GPT-Image 2.5 Flare is OpenAI's latest image generation and editing model - the newest member of the GPT-Image line that succeeded DALL-E. It continues the family's defining bet: image generation backed by reasoning. The model plans a composition before drawing it, can search the web to ground facts in the picture, and edits with the same natural-language fluency that made the GPT-Image generation an industry standard.

GPT image 2.5 Flare

Alibaba

Wan 3.0 Video

Wan 3.0 is the Alibaba Cloud Tongyi Wanxiang team's newest video generation model, and it attacks the two limits that have defined AI video for the past year: clip length and input flexibility. It generates up to 30 seconds in a single pass, and it is the first mainstream video model to accept office documents - doc, xls, ppt, pdf, and md - as generation input, not just text, images, audio, and video.

Wan 3.0 Video

Alibaba

Qwen 3.8 Flash

Qwen3.8-Flash (Flash-Next) is Alibaba's efficiency play in the 3.8 generation: a 125B-parameter multimodal MoE that activates only 6B per token, trained at one-ninth the cost of its predecessor and priced at ¥1/¥3 per million tokens. It is also the early validation platform for the architecture Alibaba is building into Qwen4.

Qwen 3.8 Flash

ByteDance

Seedance 2.0 Fast

Seedance 2.0 Fast is the speed-optimized tier of ByteDance's Seedance 2.0 family - the same unified multimodal architecture as the standard model, tuned for throughput. It generates short video clips with native synchronized audio at dramatically reduced turnaround, making it the tier for creative exploration, high-frequency iteration, and production pipelines where queue time is the bottleneck.

Seedance 2.0 Fast

Google

Veo 3.1 Lite

Veo 3.1 Lite is Google DeepMind's cost-optimized tier of the Veo 3.1 video family - the version built for developers who need to generate thousands of clips, not ten. Released on March 31, 2026, it delivers the same generation speed as Veo 3.1 Fast at more than 50% lower cost, with a per-second price under $0.10 including native audio.

Veo 3.1 Lite

Featured Models

Kling V3 0 Motion Control

Kling V3 0 Motion Control

Kling V3 0

Kling V3 0

Kling Image V3

Kling Image V3

Model API Pricing Details

happyhorse-1.1

MiniMax H3 Max

Muse Spark 1.3 Contributor

Tencent Hy4 Preview

DeepSeek V4.1 Flash

Muse Spark 1.3

Qwen Image 3.0 Pro

Grok Imagine Video 1.5

Grok Imagine Image 2.0

Gemini 3.8 Flash

Seedance 2.0 Mini

GPT image 2.5 Flare

Wan 3.0 Video

Qwen 3.8 Flash

Seedance 2.0 Fast

Veo 3.1 Lite

Kling 3

Featured Models

Kling V3 0 Motion Control

Kling V3 0 Motion Control

Kling V3 0

Kling V3 0

Kling Image V3

Kling Image V3

Model API Pricing Details

Related Models

happyhorse-1.1

MiniMax H3 Max

Muse Spark 1.3 Contributor

Tencent Hy4 Preview

DeepSeek V4.1 Flash

Muse Spark 1.3

Qwen Image 3.0 Pro

Grok Imagine Video 1.5

Grok Imagine Image 2.0

Gemini 3.8 Flash

Seedance 2.0 Mini

GPT image 2.5 Flare

Wan 3.0 Video

Qwen 3.8 Flash

Seedance 2.0 Fast

Veo 3.1 Lite