Kimi K2.5 vs Kimi K3
Compare Kimi K2.5 and Kimi K3 on key metrics including price, context length, throughput, and other model features.
AuthorMoonShotAI
Context Length262.1k
Supports Tools
Kimi K2.5 is Moonshot AI’s proprietary multimodal model, offering cutting-edge visual coding abilities and supporting a self-directed agent swarm approach. Developed from Kimi K2 and further trained on around 15 trillion mixed visual and text tokens, it achieves excellent results in general reasoning, visual coding, and autonomous tool invocation.
Activity
Last 14 daysPrompt
5MCompletion
521KTotal
5MStartup
Startup
Latency (p50)0.83s
Throughput (p50)34.1 tok/s
Pricing
Input$0.30/M tokens
Output$1.50/M tokens
Cached input$0.05/M tokens
Features
Input Modalitiestext, image
Output Modalitiestext
Supported EndpointsChat Completions
Vision
Supports Tools
AuthorMoonShotAI
Context Length1M
Supports Tools
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 navigating large repositories, using tools, debugging, and iterating against images, logs, tests, and runtime feedback. Its architecture uses KDA and Attention Residuals for computational efficiency.
Activity
Last 14 daysPrompt
7MCompletion
130KTotal
7MStartup
Startup
Latency (p50)1.97s
Throughput (p50)25.0 tok/s
Pricing
Input$1.50/M tokens
Output$7.50/M tokens
Cached input$0.15/M tokens
Features
Input Modalitiestext, image
Output Modalitiestext
Supported EndpointsChat Completions
Vision
Supports Tools