Kat Coder Air V2.5
Kat Coder Air V2.5 is the fast-response tier of KwaiPilot's KAT-Coder V2.5 pair. It keeps the context window of 256K tokens, output up to 80K tokens, and full tool surface of the Pro tier at a lower price. Your use is subject to KwaiPilot's Terms & Privacy Policies.
import { streamText } from 'ai'
const result = streamText({ model: 'kwaipilot/kat-coder-air-v2.5', prompt: 'Why is the sky blue?'})Playground
Try out Kat Coder Air V2.5 by KwaiPilot. Usage is billed to your team at API rates. Free users (those who haven't made a payment) get $5 of credits every 30 days.
Kat Coder Air V2.5
Kat Coder Air V2.5
Providers
Route requests across multiple providers. Copy a provider slug to set your preference. Visit the docs for more info. Using a provider means you agree to their terms, listed under Legal.
P50 throughput on live AI Gateway traffic, in tokens per second (TPS). Visit the docs for more info.
P50 time to first token (TTFT) on live AI Gateway traffic, in milliseconds. View the docs for more info.
Direct request success rate on AI Gateway and per-provider. Visit the docs for more info.
More models by KwaiPilot
| Model |
|---|
| 256K | 0.9s | 169tps | $0.74/M | $2.96/M | Read:$0.15/M Write:— | — | +1 |  | 07/10/2026 | ||||
| 256K | 0.9s | 114tps | $0.30/M | $1.20/M | Read:$0.06/M Write:— | — | | 03/27/2026 | |||||
| 256K | 1.8s | 107tps | $0.30/M | $1.20/M | Read:$0.06/M Write:— | — |  | 11/09/2025 |
About Kat Coder Air V2.5
Kat Coder Air V2.5 shipped alongside kat-coder-pro-v2.5 on July 10, 2026 as the fast-response option in the KAT-Coder V2.5 pair. Both models target the same job: acting inside real, executable repositories rather than generating code in a single turn. Kat Coder Air V2.5 handles issue localization, code modification, and test execution as steps in one end-to-end loop.
The capability gap between the two tiers is published rather than implied. KwaiPilot's model comparison lists a SWE-bench figure of 42.4% for Kat Coder Air V2.5 against 65.2% for kat-coder-pro-v2.5. Everything else in that comparison matches: a context window of 256K tokens, output up to 80K tokens, streaming output, context caching, MCP (Model Context Protocol), function calling, and coverage of more than 20 mainstream programming languages. KwaiPilot recommends Kat Coder Air V2.5 for fast-response agent workloads, including native OpenClaw support, and reserves the Pro tier for complex enterprise projects and SaaS integrations.
That split maps onto how coding agents actually run. Routine work dominates request volume: triaging an issue, running a test suite, applying a small patch, summarizing a diff. Send those to Kat Coder Air V2.5 and escalate to kat-coder-pro-v2.5 when a task stalls. Cached input reads bill at $0.03 per million tokens, which keeps resent repository context cheap across long sessions. Route both tiers through AI Gateway via StreamLake and switch with a model identifier change. Product details sit at .
What To Consider When Choosing a Provider
- Configuration: Kat Coder Air V2.5 and
kat-coder-pro-v2.5differ in capability, not in limits. They share a context window of 256K tokens, an output ceiling of 80K tokens, and the same tool surface. The decision rests on whether your tasks need the Pro tier's higher SWE-bench result. - Configuration: Run both against a sample of your own issues before you route production traffic. A cheaper model that loops on failed repair attempts can cost more per resolved task than a more capable one, so measure resolution rate rather than per-token price alone.
- Configuration: Context caching helps most in multi-turn agent loops that resend the same files every turn. Check the pricing panel on this page for current input, output, and cached read rates.
- 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 Air V2.5
Best for
- High-Volume Coding Agents: Request count drives the bill more than any single task
- Issue Triage and Tests: Localize a problem, run the suite, and report the result
- CI Automation: Repeated tasks of the same shape across many pull requests
- Interactive Development Loops: A window of 256K tokens holds the working set across turns
- Tiered Routing: Routine work stays here and hard tasks escalate to the Pro tier
Consider alternatives when
- Complex Repository Work:
kat-coder-pro-v2.5posts a higher SWE-bench figure on hard tasks - Multimodal Input Needed: Screenshots and diagrams require a model that accepts image input
- General-Purpose Work: Writing and open-ended analysis sit outside a coding model's scope
- Existing V2 Deployments:
kat-coder-pro-v2may already cover your context and output needs
Conclusion
Kat Coder Air V2.5 gives you the KAT-Coder V2.5 tool surface and context window at the lower price point of the pair. Use it for the routine majority of agent traffic, keep kat-coder-pro-v2.5 for the hard tasks, and route both through AI Gateway so switching tiers is a model identifier change.