Kimi K3 (batch)on Krater.ai
Kimi K3 is a 2.8T parameter open-weight multimodal reasoning model from Moonshot AI. It is suited for complex coding, knowledge work, and long-horizon agentic workflows, and is particularly strong at
Released July 16, 2026
Specifications
Kimi K3 (batch) at a glance
Context
1M tokens
Max output
943.7K tokens
Input
Text, Image, Video
Output
Text
Input price
$3.00 / 1M
Output price
$15.00 / 1M
Capabilities
What it is good at
Catalog capabilities
- Accepts image input
- Generates text responses
- Supports tool calling
- Supports structured outputs
Good for
Benchmarks
Catalog data, not Krater benchmarks.
Coding index score
76.2
Agentic index score
50.9
Intelligence index score
50.2
Design Arena webapps rank
#1
Design Arena 3d rank
#1
Compare
Compare Kimi K3 (batch)
Kimi K3 (batch) vs Kimi K2.5
Kimi K2.5 is Moonshot AI's native multimodal model, delivering state-of-the-art visual coding capability and a self-directed agent swarm paradigm.
Open comparisonKimi K3 (batch) vs Kimi K2.6
Kimi K2.6 is Moonshot AI's next-generation multimodal model, designed for long-horizon coding, coding-driven UI/UX generation, and multi-agent orchestration.
Open comparisonKimi K3 (batch) vs Kimi K2.7 Code
Kimi K2.7 Code is a coding-focused model in Moonshot AI's Kimi K2 family, built to complete end-to-end programming tasks reliably over long contexts.
Open comparisonFAQ
Questions about Kimi K3 (batch)
Clear answers based on the model catalog and Krater access.
Use Kimi K3 (batch) in your API
Connect with Krater's unified API.
Ready to use Kimi K3 (batch)?
Access leading models in one focused workspace.