MoonshotMoonshot·text

Kimi K3

CodeVisionReasoningWeb searchFunction calling
Advertised as kimi-k3
Quick reference
Kimi K3 — TLDR
  • 🧠 Ultra-large-scale open-weight reasoning model from Moonshot AI
  • 📏 Massive 1M-token context window
  • 👁️ Multimodal — reasons over images, logs, and tests
  • 🔧 Strong tool use, function calling, and web search
  • 🎯 Built for complex coding and long-horizon agentic work
  • 🌍 Open weights for self-hosting and inspection
💰 Best price on AntSeed
$0.099 / $0.44697%
per 1M · cheapest in / out
📏 Context
1M tokens
🐜 Sellers
18
advertising on AntSeed
Provider

Moonshot is an AI research lab known for developing the Kimi family of large language models. The organization has gained recognition for building capable reasoning-oriented models, with the Kimi line representing its flagship series of text generation systems.

Explore 4 more models by Moonshot
About this model

Kimi K3 is Moonshot AI's ultra-large-scale, open-weight multimodal reasoning model, built for complex coding, knowledge work, and long-horizon agentic workflows. Released in July 2026, it pairs a one-million-token context window with vision, tool use, function calling, and web search, letting it iterate against images, logs, tests, and runtime feedback rather than reasoning in isolation. Its design leans heavily toward navigating large repositories, debugging, and multi-step problem solving where a model must plan, act, and revise over extended sessions.

Within Moonshot's Kimi K lineup, K3 sits at the top of the general-purpose reasoning tier, alongside [[sibling:kimi-k2-6|Kimi K2.6]] and [[sibling:kimi-k2-5|Kimi K2.5]], while [[sibling:kimi-k2-7-code|Kimi K2.7 Code]] serves as the dedicated coding-specialized branch. As an open-weight release, it can be inspected and self-hosted, an appealing trait for teams wanting transparency alongside frontier-scale capability.

K3 is best suited to demanding agentic and engineering tasks — large-codebase reasoning, automated debugging, tool-driven research, and workflows that combine text and visual inputs across very long contexts.

View source on GitHub ↗View model card on HuggingFace ↗

This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.

Usage on AntSeed
Tokens served
3.50B
input + output
Requests
25,572
settled calls
Buyers
48
distinct, on this model
Sellers used
16
of 18 advertising
Settled
$825.68
gross USDC, this model
Sellers serving Kimi K3 (18)compare on the network explorer →
SellerReputationRoutingInput $/MCached $/MOutput $/MCategoriesAPI

"Best price" and the seller table are live AntSeed catalog data (advertised $/1M tokens — or $ per generated image for unit-billed image models — not settled amounts). Reputation = on-chain trust (0-100). "Routing" = the SDK's default buyer routing (what the VPR desktop app ships with): a trust ≥ 60 gate on the effective reputation, then cheapest-first among routable sellers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on AntSeed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.