Post-Cutoff
Search For AI agents

Moonshot AICurrentReasoning modelReleased 16 Jul 2026

Kimi K3

Moonshot’s frontier open model for long-horizon coding, knowledge work and deep reasoning; 1M context, vision.

Model idkimi-k3

Price$3 in, $15 out per 1M tokens

Checked29 September 2026

No published training cutoff

At least1,760events since its release on 16 July 2026

80 historic, 401 major. 120 days of news.

Moonshot AI doesn’t publish a training cutoff, so we count from its release date. The real gap is larger.

Five things it has never heard of

How to call it

5 ways to use Kimi K3, with the model id each provider expects.

Providers, model ids, endpoints and docs for Kimi K3
ProviderModel idEndpointDocs
Kimi API (Moonshot) kimi-k3 https://api.moonshot.ai/v1 platform.kimi.ai
Alibaba Cloud Model Studio kimi-k3 alibabacloud.com
OpenRouter moonshotai/kimi-k3 openrouter.ai
Hugging Face huggingface.co
Web app kimi.com

Pricing

Prices for Kimi K3
Input$3 per 1M tokens
Output$15 per 1M tokens
Cache read$0.30 per 1M tokens
Cache write, 5 minutes$3 per 1M tokens
Cache write, 1 hour$6 per 1M tokens

As published: “per 1M tokens (USD)”.

Source: platform.kimi.ai, checked 29 September 2026.

Specs

Input
Text, images and video
Output
Text
Context window
1,048,576 tokens
Max output
Not published
Open weights
Yes
Licence
kimi-k3 (custom)
Training cutoff
Not published
Released
16 July 2026
Status
Current
Type
Reasoning model

What stands out

What its maker marketed at launch, and what people found later, each with its source.

  • First open 3T-class model

    First

    2.8T-parameter MoE (16 of 896 experts active) - Moonshot’s claim: the first open model at this scale; weights released after launch (promised by 2026-07-27).

    Source: kimi.com

  • Kimi Delta Attention + Attention Residuals

    Hybrid linear attention (KDA) and AttnRes; ~2.5x the scaling efficiency of K2 per Moonshot.

    Source: platform.kimi.ai

  • Native vision with 1M context

    Native visual understanding (image and video) and a 1,048,576-token window; strong at coding tasks that use screenshots/visual feedback (games, frontend, CAD).

    Source: platform.kimi.ai

  • Always-on thinking with effort control

    Thinking cannot be disabled; reasoning_effort low/high/max (default max).

    Source: platform.kimi.ai

Notes

Moonshot flagship. API unlocked after a minimum $1 top-up. Chat Completions, Responses and Anthropic-compatible Messages supported. Docs moved to platform.kimi.ai (platform.moonshot.ai still serves).

curl https://api.moonshot.ai/v1/chat/completions \
 -H "Authorization: Bearer $MOONSHOT_API_KEY" -H "Content-Type: application/json" \
 -d '{"model":"kimi-k3","reasoning_effort":"high","messages":[{"role":"user","content":"Hello"}]}'

Sources: https://platform.kimi.ai/docs/guide/kimi-k3-quickstart · https://platform.kimi.ai/docs/pricing/chat · https://platform.kimi.ai/docs/models · https://www.kimi.com/blog/kimi-k3

Not yet checked against an official source: Max output and knowledge cutoff not verified.