Build long-context reasoning, coding, multimodal, and agent workflows with the Kimi models currently available through LinkModel.
Moonshot AI develops the Kimi model family. Kimi K3 is its natively multimodal, long-context model for reasoning, coding, and knowledge work, while Kimi K2.6 emphasizes agentic tasks, coding, visual understanding, and general-purpose workflows. This page only presents models that appear in LinkModel's live catalog.
Compare the Kimi language models currently listed on LinkModel. Availability, pricing, context limits, and request schemas come from the live catalog, so open a model page before production deployment.
2 models listed
Compare live Kimi models for long-context analysis, multimodal understanding, reasoning, coding, tool use, and agent workflows.
Kimi K3Moonshot AI's open flagship LLM: a 2.8T-parameter MoE (first open 3T-class model) with Kimi Delta Attention and Attention Residuals, native vision, 1M-token context, and always-on thinking for long-horizon coding, knowledge work, and reasoning.
Kimi K2.6Moonshot AI's open-weight 1T-parameter MoE multimodal agentic model, excelling in long-horizon coding, 300-sub-agent swarm orchestration, and native tool use — built for autonomous software engineering and agent workflows.
Choose Kimi K3 for native multimodal reasoning and long-context knowledge work. Choose Kimi K2.6 for agent-oriented coding and automation. Recommendations only appear when the matching model is live in LinkModel's catalog.
Moonshot presents Kimi K3 as a natively multimodal model with a 1M-token context window for long-horizon coding, deep reasoning, and knowledge work.
1M-token context · native vision · long-horizon coding · deep reasoning
Official source ↗Moonshot recommends Kimi K2.6 for agent, coding, visual-understanding, and general-purpose tasks, with Agent Swarm support in Kimi's agent product.
Use live Kimi models for long-document analysis, software engineering, multimodal reasoning, research, and agentic knowledge work from one account.
Analyze large document sets, research material, and knowledge repositories with a model selected for long-context reasoning.
Build coding assistants and multi-step engineering workflows that reason over code, call tools, and produce implementation artifacts.
Combine text and supported visual inputs for research, review, and knowledge-work tasks after confirming the exact live request schema.
Access supported Kimi models alongside other language providers while centralizing credentials, usage, and pay-as-you-go billing.
Use one LinkModel API key for supported Moonshot models and future versions as they enter the live catalog.
Evaluate Kimi alongside DeepSeek, Claude, GPT, Gemini, GLM, and other live language families without maintaining another provider billing flow.
Track requests and pay-as-you-go charges from one account while keeping the selected model explicit in each integration.
Use one LinkModel credential to call an available Kimi model and keep model access, usage, and billing in the same workflow.
Create a LinkModel API key and use it to authenticate requests for Kimi models enabled for your account.
Select Kimi K3 or K2.6 according to the workload, then open its model page to confirm the current identifier, price, context limit, and request schema.
Submit your prompt or supported multimodal input, handle the response, and monitor usage from the same LinkModel account.
2026-07-10
2026-07-22
2026-07-10
2026-07-10
FAQ
Model capabilities and selection guidance are reviewed against the official Moonshot sources listed here.
Editorial content reviewed: September 1, 2026
Moonshot and its model names belong to their respective owners. LinkModel is an independent API aggregation platform and does not imply a formal relationship.