🔥 Introducing the Helicone AI Gateway, now on the cloud with passthrough billing. Access 100+ models with 1 API and 0% markup.

moonshotai: Kimi K2 Instruct (09/05)

kimi-k2-0905

Context: 262k

Max Output: 16k

Input: $0.500/1M

Output: $2.00/1M

Enhanced version of Kimi K2 with doubled context window (256k tokens) and significantly improved coding capabilities, especially for frontend development. Features superior performance on public benchmarks (69.2% on SWE-bench Verified) and more polished code generation for web and 3D scenarios. Key Improvements 2x Context Window: Increased from 128k to 256k tokens for better long-horizon task support. Enhanced Coding: Specialized training for frontend development, UI code generation, and tool calling. Agentic Capabilities: Improved reliability for code generation rivaling frontier closed models.

Input: Text

Output: Text

Providers

deepinfra

Credits

Context262k

Max Output16k

Input$0.500/1M

Output$2.00/1M

Cache Read$0.400/1M

Cache Write—

novita

Credits

Context262k

Max Output16k

Input$0.600/1M

Output$2.50/1M

Cache Read—

Cache Write—

groq

Credits

Context262k

Max Output16k

Input$1.00/1M

Output$3.00/1M

Cache Read$0.500/1M

Cache Write—

openrouter

Credits

Context262k

Max Output16k

Input (Max)$1.42/1M

Output (Max)$5.28/1M

Cache Read—

Cache Write—

Quick Start

Use Kimi K2 Instruct (09/05) through Helicone's AI Gateway with automatic logging and monitoring.