KAT-Coder-Pro V2
KAT-Coder-Pro V2 by KwaiKAT is an agentic coding model with 256K context, tool use, structured outputs, and enterprise software engineering capabilities.
Model Overview
Capabilities, design details, and architectural traits
KAT-Coder-Pro V2 - Agentic Coding Model
KAT-Coder-Pro V2 is KwaiKAT's flagship agentic coding model designed for enterprise software engineering, large-scale codebases, SaaS integration, and multi-step developer workflows.
Defining Characteristics
- Agentic coding focus: Built for complex software engineering tasks including multi-file editing, issue resolution, system integration, and long-running coding workflows.
- 256K context window: Supports large codebases and extended development sessions through a long-context architecture.
- Structured tool workflows: Supports tool use, function calling, structured outputs, and agent-based development pipelines.
- Specialize-then-Unify training: Uses specialized training domains including SWE, WebCoding, Terminal, WebSearch, and General before combining capabilities into a unified coding model.
Core Design
KAT-Coder-Pro V2 focuses on production-grade software engineering. It also supports web aesthetics generation for creating polished landing pages and presentation materials.
Benchmark Performance
Independent evaluations · Artificial Analysis
Accuracy & Capability Details
Compare Models Side-by-Side
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | KwaiKAT | Anthropic | Anthropic |
| Release Date | March 27, 2026 | September 22, 2026 | September 28, 2026 |
| Knowledge Cutoff | - | - | Jun 2026 |
| Context & Limits | |||
| Context Window | 256K | 1M Best Context Window | 1M Best Context Window |
| Pricing (per 1M tokens) | |||
| Input Pricing | $0.30 Best Input Pricing | $4 | $2 |
| Output Pricing | $1.20 Best Output Pricing | $20 | $10 |
| Modalities | |||
| Inputs | text | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 21.7 | 57.6 Best Intelligence Index | 56.0 |
| Coding Index | 59.5 | - | - |
| Agentic Index | 15.1 | - | - |
Humanity's Last Exam
Extremely difficult logical reasoning and knowledge.
Long Context Reasoning
Logical reasoning over long context windows.
SciCode Benchmark
Scientific coding and mathematical modeling.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
Top AI Models
Leading alternatives by intelligence score
