Kimi K3, for deep reasoning
Kimi K3 from Moonshot AI thinks through hard problems step by step before it answers. In SelGPT it sits next to GPT-6.1, Claude and Gemini.
By the SelGPT team. Updated
Kimi K3 is a reasoning model from Moonshot AI that works through hard, multi-step problems before answering. SelGPT includes it in Plus from $9.99 a month billed yearly, served only by hosts that agree not to train on your chats.
What it is good at
- Maths, logic and planning problems with many steps.
- Comparing options and recommending one, with the reasons.
- Agent-style tasks that use tools such as web search.
Plans
It draws from your plan’s monthly allowance, shown as a percentage in the app. GPT-6 Luna stays unlimited on paid plans when the allowance runs out.
Questions
Who makes Kimi K3?
Moonshot AI. In SelGPT it runs on hosts that agree not to train on your requests.
Related
- AI for researchResearch with SelGPT: it searches the web, pulls several sources together, compares them and shows where every fact came from.
- AI for mathSolve and understand maths problems step by step with SelGPT’s reasoning model, from algebra and statistics to proofs and word problems.
- ModelsEvery model in SelGPT: GPT-6 Luna, Claude Opus 5.5, GPT-6.1 Sol, Gemini 3.8 Flash, Kimi K3, DeepSeek, MiMo, GLM and Qwen, in one chat.
- GPT-6 LunaUse OpenAI’s GPT-6 Luna in SelGPT for everyday questions, writing, files and images, unlimited within fair use on every paid plan.
- Claude Opus 5.5Use Anthropic’s Claude Opus 5.5, ranked #1 on the Artificial Analysis Intelligence Index, in SelGPT Plus from $9.99 a month billed yearly.
- GPT-6.1 SolUse OpenAI’s GPT-6.1 Sol in SelGPT for writing, analysis and code, alongside Claude, Gemini and Kimi, from $9.99 a month billed yearly.