Skip to content

KAT-Coder-Pro V1

View Status

KAT-Coder-Pro V1 is KwaiPilot's agentic coding model. It achieves a 73.4% resolve rate on SWE-Bench Verified with a context window of 256K tokens, parallel tool calling, and multi-turn support.

Reasoning
index.ts
import { streamText } from 'ai'


const result = streamText({
  model: 'kwaipilot/kat-coder-pro-v1',
  prompt: 'Why is the sky blue?'
})

What To Consider When Choosing a Provider

  • Configuration: KAT-Coder-Pro V1 targets multi-turn agentic sessions. Factor context window consumption and session state into your integration planning. See https://novita.ai for methodology and benchmark details from KwaiPilot.
  • Zero Data Retention: AI Gateway does not currently support Zero Data Retention for this model. See the documentation for models that support ZDR.
  • Authentication: AI Gateway authenticates requests using an API key or OIDC token. You do not need to manage provider credentials directly.

When to Use KAT-Coder-Pro V1

Best For

  • Automated issue resolution: Pull request generation in real repositories end-to-end
  • Multi-file refactoring: Refactoring workflows that span many files in one session
  • Parallel tool pipelines: Agent pipelines where parallel tool calling reduces total completion time
  • Scale test generation: Test case generation across existing codebases

Consider Alternatives When

  • General reasoning needs: Writing or multimodal input sits outside a coding-tuned model's scope
  • Simple completions: A lighter model suffices with minimal context
  • LiveCodeBench benchmark: Competitive programming is the primary evaluation criterion

Conclusion

KAT-Coder-Pro V1 pairs a 73.4% SWE-Bench Verified resolve rate with parallel tool calling and multi-turn support. Use it for software engineering automation across real repository tasks. Route requests through AI Gateway for access via novita.

Frequently Asked Questions

  • What is KAT-Coder-Pro V1's SWE-Bench Verified score?

    73.4% resolve rate on SWE-Bench Verified, which tests autonomous resolution of real GitHub issues.

  • What does parallel tool calling mean in practice?

    The model issues multiple tool calls in a single inference step instead of waiting for each response sequentially. That cuts latency when several operations can run at once.

  • What is the context window size?

    KAT-Coder-Pro V1 has a context window of 256K tokens. You can fit large codebases, multi-file diffs, and extended conversation history in a single context.

  • What types of software engineering tasks does it support?

    Eight task types: feature implementation, feature enhancement, bug fixing, refactoring, performance optimization, test case generation, code understanding, and configuration and deployment.

  • What is the pricing for KAT-Coder-Pro V1?

    Current pricing is shown on this page. AI Gateway routes across providers, and rates may vary by provider.

  • How do I try KAT-Coder-Pro V1?

    Open the playground at https://ai-sdk.dev/playground/novita:kwaipilot/kat-coder-pro or call the model through AI Gateway with your provider credentials.