All AI models · LLM benchmarks · Methodology

Moonshot AI

Kimi K3

Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work

Model facts

Context window
1.05M tokens
Maximum output
131K tokens
Cheapest paid input
$2 per 1M tokens
Cheapest paid output
$8 per 1M tokens
Open weights
yes
Providers
71

Capabilities and modalities

reasoning, tool calling, structured output, attachments, temperature control, open weights, text, image, video, pdf.

Local hardware estimate

At 4K context: Q4_K_M 1702.4 GiB, Q8_0 2996.9 GiB, F16 5424.1 GiB working memory.

Parameters
2779.9 billion
Architecture
kimik3
Model license
other

Planning estimate adapted from the MIT-licensed whichllm estimator using metadata from Hugging Face; not a vendor minimum requirement.

Artificial Analysis benchmarks

Compare this model on the full LLM benchmark leaderboard.

ConfigurationAAICodingMathOutput tokens/s
max59.776.2not measured37.9
low48.372.0not measured33.7

Observed price history

  • 2026-08-24: $2 input / $8 output per 1M tokens

API providers