Alibaba
Open weights

Kimi K3

Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work

Intelligence
AA Index
Output price
$15/1M
In $3/1M
Context
1049k
tokens
Speed
tokens/sec

Performance

Benchmark data for this model isn't available yet.

Specifications

Provider
Alibaba
Family
kimi-k3
Released
2026-07-16
Context window
1,048,576 tokens
Max output
1,048,576 tokens
Input price
$3/1M
Output price
$15/1M
Cache read
$0.3/1M
Input modalities
text
image
Output modalities
text
Capabilities
Reasoning · Tool calling · Structured output · File attachments · Open weights

Specs from models.dev · benchmarks from Artificial Analysis.

AI tools built on Alibaba

Explore directory tools powered by Alibaba models.

Vynaris logo
Vynaris
Generic

Vynaris hosts powerful uncensored models—including Qwen3.8-27B, DeepSeek-V4-Flash-0731, and Qwen3.6-35B-A3B—through an OpenAI-compatible API for authorized security testing and research.

Wan 3.0 AI Video Generator logo

Wan 3.0, powered by Alibaba Cloud, is an all-in-one AI video generator that creates cinematic videos up to 30 seconds from diverse inputs. Transform text prompts, images, or multimodal references into stunning 1080P visuals with integrated native audio. Guide your scenes precisely with first and last frame controls, maintain consistency across shots, and choose from landscape, square, or vertical formats. It's an efficient solution for crafting professional-grade product ads, social media content, and short cinematic narratives, simplifying complex video production for a wide range of creative needs.

Wan 3 Video logo
Wan 3 Video
Video Generation

Unlock cinematic video creation with Wan 3 Video, leveraging Alibaba Cloud Model Studio's advanced AI. Transform text into stunning visuals, convert first frames into dynamic shots, control transitions with precision, and generate performance videos from references. Designed for every creator, from influencers to product teams, to produce high-quality videos 30x faster, with multi-shot narrative capabilities and 720P/1080P output. Experience versatile AI video generation with unparalleled control and cinematic quality.

Qwen Image Edit logo
Qwen Image Edit
Image Generation

Qwen Image Edit is an advanced AI-powered tool that delivers effortless, text-driven image editing with unmatched precision and creativity. Leveraging a 20B parameter model, it transforms your ideas into stunning visuals in seconds, offering semantic, appearance, and bilingual text modifications. Achieve professional-grade results for virtual avatar creation, background replacement, style transfer, and fine-grained detail refinement, all while maintaining visual consistency. Perfect for creators and professionals seeking high-quality AI image generation and editing.