Kimi
Evaluate access, provider, deployment and enterprise considerations from one stable model page.
- Access
- Providers
- Coding
Chinese AI topic hub
Compare leading Chinese AI models while keeping APIs, pricing, providers, self-hosting and enterprise requirements visible in the same decision.
Current verified update
DECLAREDKimi's official materials now identify Kimi K3 as its flagship model, with managed API and open-weight routes. Other model pages remain stable while their current details are verified.
Research baseline
July 28, 2026
Availability, pricing and provider routes can change. Follow the source attached to each detail.
Model entities
The topic hub connects the ecosystem; each model keeps a stable URL under /models/.
Evaluate access, provider, deployment and enterprise considerations from one stable model page.
Review API, self-hosting, provider and enterprise paths without mixing the model with its delivery route.
Explore access, provider, self-hosting and enterprise considerations for a global deployment decision.
Assess API, coding, self-hosting and enterprise paths from a stable model-level overview.
Start from your task
Start with stable entity pages for Kimi, Qwen, DeepSeek and GLM rather than a news-driven list.
Browse model guides →Compare official platforms, API keys and third-party provider routes without treating them as identical.
Start with Kimi API →Keep the original billing currency, cache rules, provider and region attached to every quoted price.
Review Kimi pricing →Compare the coding client, model, provider, open-weight route and data-control path separately.
Compare coding agents →Open weights move control toward your infrastructure, but also move operations and model-serving work to your team.
Open deployment guides →Bring provider identity, region, retention, contracts and procurement evidence into the adoption decision.
Evaluate for enterprise →Evaluation model
A model can be open-weight while a managed API still sends data through an external provider. A familiar model name does not answer the data-path question.
01
The weights and capabilities you intend to use.
02
The company serving requests or routing them onward.
03
Official API, cloud provider, aggregator, VPC or self-hosted runtime.
04
Where prompts, files, responses, logs and credentials actually travel.