FitMyLLM
Moonshot/Dense

MKimi-K2 1026.5B

- Messages with name field are now supported. We’ve also moved the chat template to a standalone file for easier viewing.

chat
1026.5B
Parameters
128K
Context length
16
Benchmarks
17
Quantizations
95K
HF downloads
Architecture
Dense
Released
2025-07-11
Layers
61
KV Heads
64
Head Dim
112
Family
kimi

Quantization Options

Context length:
QuantBitsVRAM @ 16KQuality
IQ2_XXS2.38
306.7 GB
305.9 + 0.8 KV
low
IQ2_M2.93
377.2 GB
376.4 + 0.8 KV
low
Q2_K3.16
406.8 GB
406.0 + 0.8 KV
low
IQ3_XXS3.25
418.3 GB
417.5 + 0.8 KV
low
IQ3_XS3.5
450.4 GB
449.6 + 0.8 KV
low
Q3_K_S3.64
468.4 GB
467.5 + 0.8 KV
low
IQ3_M3.76
483.7 GB
482.9 + 0.8 KV
low
Q3_K_M4
514.5 GB
513.7 + 0.8 KV
low
Q3_K_L4.3
553.0 GB
552.2 + 0.8 KV
moderate
IQ4_XS4.46
573.6 GB
572.8 + 0.8 KV
moderate
Q4_K_S4.67
600.5 GB
599.7 + 0.8 KV
moderate
Q4_K_M4.89
628.7 GB
627.9 + 0.8 KV
good
Q5_K_S5.57
716.0 GB
715.2 + 0.8 KV
good
Q5_K_M5.7
732.7 GB
731.9 + 0.8 KV
good
Q6_K6.56
843.0 GB
842.2 + 0.8 KV
excellent
Q8_08.5
1091.9 GB
1091.1 + 0.8 KV
lossless
FP1616
2054.3 GB
2053.5 + 0.8 KV
lossless

Select your GPU above to see speed estimates and compatibility for each quantization.

Too big for a single GPU — plan a multi-GPU deployment
Even the lightest quant needs ~307 GB. Size GPUs, replicas, TCO and scaling for a production setup. Open in Enterprise →
▸ READY TO RUN THIS?RENT BY THE HOUR

RENT A GPU AND RUN KIMI-K2 1026.5B NOW

Rent on Vast.ai →Or RunPod →

Spin up an A100 / H100 / 4090 in ~60s. Pay by the second. Cancel anytime.

Community Ratings

Loading ratings...

Benchmarks (16)

MATH-50097.4
IFEval89.8
MMLU-PRO81.1
GPQA Diamond75.1
SWE-bench65.8
τ²-Bench61.1
AA Math57.0
LiveCodeBench53.7
AA Long Context51.0
AIME49.5
IFBench41.5
SciCode34.5
AA Intelligence26.3
AA Coding22.1
Terminal-Bench15.9
HLE4.7

Run this model

▸Easiest way to get started·Beginners
DOCS ↗
curl -fsSL https://ollama.com/install.sh | sh
$ollama run kimi:1027b-q4_K_M

Tag may need adjustment — check ollama.com/library/kimi for available tags.

▸ SETUP GUIDE
>_

Auto-setup with fitmyllm CLI

Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.

pip install fitmyllmthen run fitmyllmLearn more
Auto-detect GPULive tok/s in chatSpeed benchmarks9 inference engines
HuggingFaceOllama LibraryGGUF DownloadsBuild Hardware

Find the best GPU for Kimi-K2 1026.5B

Build Hardware for Kimi-K2 1026.5B
▸ SPEC SHEET

Kimi-K2 1026.5B — 1026.5B Dense.

▸ SPECIFICATIONS
PARAMETERS
1026.5B
ARCHITECTURE
Dense Transformer
CONTEXT LENGTH
128K tokens
CAPABILITIES
chat
RELEASE DATE
2025-07-11
PROVIDER
Moonshot
FAMILY
kimi
▸ VRAM REQUIREMENTS
QUANTBPWVRAMQUALITY
IQ2_XXS2.38305.9 GB65%
IQ2_M2.93376.4 GB75%
Q2_K3.16406.0 GB78%
IQ3_XXS3.25417.5 GB82%
IQ3_XS3.5449.6 GB84%
Q3_K_S3.64467.5 GB85%
IQ3_M3.76482.9 GB86%
Q3_K_M4513.7 GB88%
Q3_K_L4.3552.2 GB90%
IQ4_XS4.46572.8 GB92%
Q4_K_S4.67599.7 GB93%
Q4_K_M4.89627.9 GB94%
Q5_K_S5.57715.2 GB96%
Q5_K_M5.7731.9 GB96%
Q6_K6.56842.2 GB97%
Q8_08.51091.1 GB100%
FP16162053.5 GB100%
§ 01BENCHMARK SCORES
MMLU-PRO81.1
IFEval89.8
MATH-50097.4
GPQA Diamond75.1
LiveCodeBench53.7
SWE-bench65.8
AIME49.5
HLE4.7
aa_ifbench41.5
aa_terminal_bench15.9
aa_tau261.1
aa_scicode34.5
aa_lcr51.0
AA Intelligence26.3
AA Coding22.1
AA Math57.0
§ 02RUN COMMAND

Run Kimi-K2 1026.5B locally with Ollama — needs 627.9 GB VRAM at Q4_K_M:

$ollama run kimi:1026b