Sovereign inference — Kimi K3 Fast

Your sovereign, private Kimi K3 Fast

Kimi K3 is Moonshot AI's flagship open-weight, natively multimodal agentic model: a 2.8T-parameter MoE (104B active) built on Kimi Delta Attention and Attention Residuals, with native vision, a 1M-token context window and interleaved thinking with configurable reasoning effort. It is designed for frontier intelligence across long-horizon coding, deep research and agentic knowledge work, sustaining multi-hour engineering sessions and orchestrating terminal tools with minimal oversight. Served on the fast service tier. A reasoning model in the ultra tier on the fast serving route, served on ToothFairyAI's AU, EU and US endpoints — private by default, billed in US dollars per million tokens at the published rates below. No training on your data, no vendor lock-in.

AU · EU · US resident$6.30/1M input$31.50/1M output$0.63/1M cached input800K contextFast serving

How Kimi K3 Fast compares

Independent intelligence index for Kimi K3 Fast against every peer in the Ultra tier (and the tier above where needed) — same source, same tests.

Kimi K3 Fast (this model)43.6Kimi K3.1 Pro High Speed46.0Qwen3.8 Max45.4Qwen 3.8 2.4T A95B39.9

Throughput — tokens per second, median

Kimi K3 Fast37.3 t/sKimi K3.1 Pro High Speed39.1 t/s

Latency — end-to-end seconds for a 500-token answer

Kimi K3 Fast71.2sKimi K3.1 Pro High Speed66.7sQwen3.8 Max67.1s

Capability indexes

Domain-weighted agentic performance for Kimi K3 Fast.

Finance & AccountingStrategy & OpsLegalHealthcare & MedicalEngineeringEconomics

Capability index vs ultra peers

13.326.740.053.446.950.548.645.043.353.4Kimi K3 Fast45.647.346.141.446.452.3Qwen3.8 Max42.644.041.736.040.349.0Qwen 3.8 2.4T A95B
Finance & AccountingStrategy & OpsLegalHealthcare & MedicalEngineeringEconomics

The evaluations behind the index

Independent benchmark results for Kimi K3 Fast — Elo scores as published; pass rates shown as percentages.

AutomationBench-AA58.3 %Terminal-Bench 4.012.6 %SciCode59.5 %Humanity's Last Exam46.9 %GDP.pdf22.0 %CritPt23.4 %AA-LCR (long context)88.7 %

Per-evaluation economics

EvaluationScoreTime / taskOutput tokens / task
AA-Briefcase1504.0449.5 min119873
GDPval-AA1523.9724.5 min59382
AutomationBench-AA58.3%8.2 min19920
Terminal-Bench 4.012.6%50.8 min122899
SciCode59.5%1.5 min3520
Humanity's Last Exam46.9%10.8 min26162
GDP.pdf22.0%4.0 min9597
CritPt23.4%24.5 min59238
AA-Omniscience19.703.6 min8813
AA-LCR (long context)88.7%0.6 min1529

Run Kimi K3 Fast privately

One key, one endpoint, three jurisdictions.

  • Size tierUltra
  • Serving routeFast
  • Input rate$6.30/1M
  • Output rate$31.50/1M
  • Cached input rate$0.63/1M
  • Context window800K tokens
  • Max output120K tokens
  • ReasoningYes — emits thinking traces
  • Vision & video inputSupports image input
  • Tool callingSupported
  • Data residencyAU, EU and US endpoints — processing stays inside the region you call