KAT Models by Kwaipilot
KwaipilotProprietary
5 models2025–2026Up to 256k ctxFrom $0.03/1M input
Details
ResearcherKwaipilot
LicenseProprietary
Commercial useCommercial use: conditional
Models5
Released2025–2026
Max context256k
Capabilities
Function Calling2 of 5 models
Tool Use2 of 5 models
Structured Outputs1 of 5 models
Code Execution1 of 5 models
About
The KAT (Kwai AutoThink) family of models from Kwaipilot (Kuaishou's AI-for-SE division), including open-source agentic coding models (KAT-Dev series) and proprietary high-performance coding models (KAT-Coder-Pro series).
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
5 in view
KAT Coder Pro V2Current
Use when the workload needs code, 256k context, and tool use.
2026-03code256k contexttool use
KAT Coder Pro V1Current
Use when the workload needs code, 256k context, and tool use.
2025-09code256k contexttool use
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| KAT Coder Pro V2 | Use when the workload needs code, 256k context, and tool use. | 2026-03 | code256k contexttool use | Current |
| KAT Dev 32B | Use when the workload needs code and 32B parameters. | 2025-10 | code32B parameters | Current |
| KAT Dev 72B Exp | Use when the workload needs code and 72B parameters. | 2025-10 | code72B parameters | Current |
| KAT Coder | Use when the workload needs code and 256k context. | 2025-09 | code256k context | Current |
| KAT Coder Pro V1 | Use when the workload needs code, 256k context, and tool use. | 2025-09 | code256k contexttool use | Current |
Release Timeline
3 release groups2026-03
1 current
KAT Coder Pro V2
Currentcode256k contexttool use
2025-10
2 current
KAT Dev 32B
Currentcode32B parameters
KAT Dev 72B Exp
Currentcode72B parameters
2025-09
2 current
KAT Coder
Currentcode256k context
KAT Coder Pro V1
Currentcode256k contexttool use
Specifications(5 models)
| Model | Released | Context | Parameters | Fn Calling | Tool Use | Structured Outputs | Code Exec |
|---|---|---|---|---|---|---|---|
| KAT Coder Pro V2 | 2026-03 | 256k | — | Yes | Yes | Yes | Yes |
| KAT Dev 32B | 2025-10 | — | 32B | No | No | No | No |
| KAT Dev 72B Exp | 2025-10 | — | 72B | No | No | No | No |
| KAT Coder | 2025-09 | 256k | — | No | No | No | No |
| KAT Coder Pro V1 | 2025-09 | 256k | — | Yes | Yes | No | No |
Available From(4 providers)
Pricing
| Model | Provider | Input / 1M | Output / 1M | Type |
|---|---|---|---|---|
| KAT Coder Pro V1 | Vercel AI Gateway | $0.03 | $1.2 | Serverless |
| KAT Coder Pro V2 | OpenRouter | $0.3 | $1.2 | Serverless |
| KAT Coder Pro V2 | Vercel AI Gateway | $0.3 | $1.2 | Serverless |
| KAT Coder Pro V2 | Novita AI | $0.3 | $1.2 | Serverless |
| KAT Coder | Fireworks AI | $0.9 | $0.9 | Serverless |
| KAT Dev 32B | Fireworks AI | $0.9 | $0.9 | Serverless |
| KAT Dev 72B Exp | Fireworks AI | $0.9 | $0.9 | Serverless |
Frequently Asked Questions
- What is KAT used for?
- KAT is used for code, agent workflows and tool use, and structured outputs. The family description and listed model capabilities point to those workloads as the best fit.
- How does KAT compare to Claude 3?
- KAT by Kwaipilot is strongest where you need code, while Claude 3 by Anthropic is the closest related family to check for vision and multimodal work. KAT has 5 listed variants and reaches up to 256k context, while Claude 3 reaches up to 200k context, so compare the specs and pricing tables before choosing a production model.
- Which KAT model should I use?
- For the lowest listed input price, start with KAT Coder Pro V1 through Vercel AI Gateway at $0.03/1M input tokens. For the most capable/latest local choice, evaluate KAT Coder Pro V2 with 256k context and tool use, function calling, and structured outputs.