Skip to content
Model Pareto

Kimi-K3-INT4

Red Hat AIOpen weightsINT4quantized from Kimi K3huggingface.co ↗
Compare with nearest rivalsCompare with…
Input
—
Output
—
Speed
~67tok/sest.
Context
1Mtokens

Rankings

Self-hostingEstimated

How we estimate →

What it would cost to run these open weights yourself on rented GPUs. No API sells this model, so this is its price on the chart.

Hardware
16× B200 180GB
INT4 weights
Throughput
~21,178 tok/s
many requests batched
Price
$0.61–5.44 /M tok
busy → light use
On your own machine
Too large
needs data-center GPUs
Assumptions (5)
  • INT4 weights (2800B params, 104B active per token) + 40% KV-cache headroom ≈ 1960 GB
  • 16× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
  • ~21,178 output tok/s aggregate at batch 713 (bandwidth-bound); MoE compute scales with active params
  • Blended 3:1 input:output; prefill ~139,024 tok/s
  • Low = 75% utilization at the low GPU price; high = 20% at the high price

Benchmark scores

~ italic, dashed = no published score yet, estimated from related benchmarks.

Hard reasoning

Humanity's Last ExamEstimated
~#46 of 185 (estimated)~29.7%
  1. #1Claude Opus 5.561.4%
  2. #2Claude Fable 5.159.1%
  3. #45Inkling31.9%
  4. ~#46Kimi-K3-INT4Estimated29.7%
  5. #46A.X-K229.6%
No published score — estimated from related benchmarks
AIME (latest)Estimated
~#2 of 32 (estimated)~98.9%
  1. #1GLM-5.3 NVFP499.5%
  2. ~#2Kimi-K3-INT4Estimated98.9%
  3. #2Inkling97.1%
No published score — estimated from related benchmarks
AA Intelligence IndexEstimated
~#29 of 86 (estimated)~32.3
  1. #1Claude Opus 5.557.6
  2. #2Claude Sonnet 5.556.0
  3. #28Qwen3.8 27B33.7
  4. ~#29Kimi-K3-INT4Estimated32.3
  5. #29K2 Horizon 375B A23B30.5
No published score — estimated from related benchmarks