ModelPriceWatch.com
Last scan 2026-08-30 Models tracked 237 Providers 31 Cheapest paid Granite 4.0 H Micro $0.017/Mtok in Every price links to its source

Kimi K3

by Moonshot

Current new · 45d flagship open weights mid tier
Email me when Kimi K3 gets cheaper

Today's price · per 1M tokens

Input

$3.00

per 1M tokens

Output

$15.00

per 1M tokens

Blended

$6.00

blended $/1M — 3:1 weighted input:output

Cached input

$0.300

10% of input — prompt caching

Source: official Moonshot pricing · read Jul 16, 2026 MODELPRICEWATCH.COM · 2026-08-30

Price receipt

We read Moonshot's own pricing page on Aug 24, 2026 and found kimi-k3 listed with a price on the same row — that read is where the number above comes from. We keep the page text we read; its content hash is aef6cfa074.

Overview

Moonshot's flagship 2.8T-parameter open-weight (MoE) multimodal reasoning model with a 1M-token context, built for complex coding, knowledge work, and long-horizon agentic workflows. OpenAI-compatible API. $3.00/$15.00 per 1M tokens; cached $0.30.

Run it yourself

Deploy this open model on rented GPUs

Open weights mean you can self-host instead of paying per-token API prices. These platforms let you serve it on demand — often cheaper at scale.

RunPod →
Rent GPUs by the second and serve any open-weight model yourself.
Novita AI →
Serverless open-model inference billed per token — no GPU ops.
Vast.ai →
GPU marketplace — rent spot/on-demand GPUs cheaply and serve any open-weight model.

Partner links — we may earn a commission if you sign up, at no cost to you. We only list platforms we'd recommend regardless.

Capabilities

struck through = not supported
Input 2/5
Text ✓ Image ✓ Audio Video PDF
Output 1/5
Text ✓ Image Audio Video Embedding
Features 5/9
Prompt caching ✓ Reasoning ✓ Coding Fast inference Long context ✓ Open weights ✓ Multimodal ✓ Web search Realtime

Benchmark performance

accuracy % · higher is better
Avg benchmark score83.3%
Perf per $/Mtok13.9
GPQA Diamond
93.5%
SWE-Bench Verified
93.4%
Terminal-Bench 2.1
88.3%
MCP Atlas
84.2%
BrowseComp
91.2%
Humanity's Last Exam
56%
HumanEval
76.8%
Independent composite scores
  • AA Intelligence Index: 59.7 (Artificial Analysis composite across reasoning, knowledge and coding evals)
  • AA Agentic Index: 54.3 (Tool use, planning, autonomy)
  • AA-Omniscience Index: 19.7 (−100–100)
  • GDPval-AA v2: 1643.7 ELO (Real-world work tasks, human baseline = 1000)
  • LMSYS Chatbot Arena: 1485.3 ELO (Human preference)
How it stacks up
  • Ranks #8 of 73 benchmarked models by average score
  • Ranks #11 of 169 comparably-measured models by percentile score, across 12 independent measurements
  • Ranks #135 of 169 comparably-tested models by normalized performance per dollar
  • Strongest at MCP Atlas — 84.2%, #1 of 28

35.2 tokens/sec output 4.46s latency to first token (TTFT)

Source: Vellum LLM Leaderboard · updated Aug 30, 2026 · See full rankings →

Specifications

Provider
Moonshot
Context window
1M tokens
Modality
text, image
Parameters
Proprietary
Open source
Yes — open weights available
Released
Jul 16, 2026
Status
Current
Last updated
Jul 16, 2026
Tags
open-weightsflagshipcachingmultimodalreasoning

Availability verified: Aug 24, 2026 — listed on Moonshot's own page

Try on Moonshot Official pricing page