API pricing per million tokens
- Input
- $1.4
- per 1M, InferenceNet (fp4)
- Output
- $10.8
- per 1M tokens
- Cached input
- $0.3
- no batch tier listed
- Context
- 1M
- 944K max output
Where to buy it (20)
- InferenceNetfp4, fp4$1.4 / $10.8cached $0.3
- Sail Researchfp4, fp4$1.5 / $11.5cached $0.3
- Phaladefault region$2.1 / $10.5cached $0.21
- Relacefp4, fp4$2.1 / $10.5cached $0.21
- Makoradefault region$2.55 / $12.8cached $0.256
- Waferdefault region$1.99 / $14.5cached $0.199
DigitalOceandefault region$2.55 / $13cached $0.255
- Morphdefault region$2.5 / $14cached $0.29
DeepInfrabf16, bf16$2.85 / $14.3cached $0.285
Basetenfp8, fp8$3 / $15cached $0.3
Chutesmxfp4, mxfp4$3 / $15cached $0.3
Fireworks AIdefault region$3 / $15cached $0.3
Modalmxfp4, mxfp4$3 / $15cached $0.3
Moonshot AImxfp4, mxfp4$3 / $15cached $0.3
Parasailfp4, fp4$3 / $15cached $0.3
Together AIdefault region$3 / $15cached $0.3
Qwen (Alibaba)default region$3.45 / $17.3cached $0.345
- Morphfast$6 / $22.5cached $0.6
Through a gateway
Your own numbers100M input and 20M output tokens a month: $355.00 direct.
Helicone$355.00+0%
Kong AI Gateway$355.00+0%
LiteLLM$355.00+0%
Vercel AI Gateway$355.00+0%
AWS Bedrock$355.00+0%
Azure AI Foundry$355.00+0%
Google Vertex AI$355.00+0%
Databricks Mosaic AI$355.00+0%
Hugging Face Inference$355.00+0%
Cloudflare AI Gateway$372.75+5%
Requesty$372.75+5%
Eden AI$374.52+5.5%
OpenRouter$374.52+5.5%
Snowflake Cortex AI$426.00+20%
Martian Gatewayn/a
Portkeyn/a
TrueFoundry AI Gatewayn/a
IBM watsonx.ain/a
OCI Generative AIn/a
NVIDIA NIMn/a
Related models
GPT-5.6 SolOpenAI · 1.05M$2 / $10in / out per 1M$2$10$0.21.05MJul 9, 2026
Muse Spark 1.3Meta · 1M$1.25 / $4.25in / out per 1M$1.25$4.25$0.151MSep 2, 2026
Claude Opus 5.5Anthropic · 1M$4 / $20in / out per 1M$4$20$0.21MSep 22, 2026
Gemini 3.1 Pro PreviewGoogle Gemini API · 1M$2 / $12in / out per 1M$2$12$0.21MFeb 19, 2026
About Kimi K3
Kimi K3 is a 2.8T parameter open-weight multimodal reasoning model from Moonshot AI.
Takes text, image, video in and returns text. First listed Sep 23, 2026. Blended price $3.74 per 1M tokens at 3 input to 1 output. Overall score 11 on leadersboard.fru.dev (rank 18).
Questions
How much does Kimi K3 cost?
$1.4 per million input tokens and $10.8 per million output tokens, with cached input at $0.3.
What is the context window of Kimi K3?
1M tokens, with up to 944K tokens of output.
Where is Kimi K3 cheapest?
Of 20 routes tracked, InferenceNet is cheapest at $1.4 input and $10.8 output per million tokens.
Has Kimi K3's price changed?
Yes. The input price went from $2.49 to $1.99, first seen Sep 24, 2026.
List prices from providers' pages on the date shown; your contract may differ. Logos via logo.dev; trademarks belong to their owners.