Kimi K3
Q4 GGUF download: mradermacher/Qwen3.5-9B-Kimi-k3-Distilled-GGUF - 5.78 GB on disk, read 2026-10-26. VRAM figures above are estimates; this is the actual file size.
Intelligence benchmarks
Artificial Analysis indexes - compared with the best open and proprietary models
Intelligence
17.4
AA Index
Intelligence Index - Kimi K3 vs. the field
Best open-weight models (you can run locally) and leading proprietary models for context.
Benchmark data from Artificial Analysis · updated 2026-10-31.
Hosted API cost & speed
What it costs to rent this model instead of running it locally
| Metric | Value |
|---|---|
| Input (per 1M tokens) | $3.00 |
| Output (per 1M tokens) | $15.00 |
| Median speed (hosted API) | 58 tok/s |
Median across hosted providers, measured by Artificial Analysis - updated 2026-10-31. Local llama.cpp speed depends on your hardware — see the hardware pages for estimates.
Arena rating (LMArena)
Human preference wins from head-to-head chat battles
| Metric | Value |
|---|---|
| Overall arena rating | 1475.7 |
| Rank (overall) | #25 |
| Battle votes | 36,780 |
Overall rating published by LMArena on 2026-10-28 - a pairwise human-preference Elo across live chat battles; higher is better. Fetched 2026-10-31. Open-weights models compete head-to-head with proprietary ones here — no benchmark prompts, just what people pick.
Standard benchmarks
Performance across standard evaluations
| Benchmark | Score |
|---|---|
| GPQA Diamond | 93.5 |
| HLE | 46.9 |
| SciCode | 59.5 |
Source: Artificial Analysis component evals - updated 2026-10-31.
Will it run on your hardware?
Pick your GPU memory - see which quantizations fit, and the cheapest card for the rest
Need an exact figure for your context length? Use the VRAM calculator.
Run it locally
Copy-paste - running in under a minute
vllm serve moonshotai/Kimi-K3New to this? Start with Ollama · serve to many users with vLLM.
Deep dive
Notes, sources, and the full write-up
Kimi K3 is a 2779.9B-parameter other model from Kimi. It scores 17.4 on the Artificial Analysis Intelligence Index. At Q4_K_M it needs roughly 1612 GB of VRAM, placing it in the 48 GB+ / multi-GPU hardware tier.
Benchmarks
Artificial Analysis Intelligence Index - Kimi K3 vs. leading closed models:
| Model | Intelligence | Coding | GPQA |
|---|---|---|---|
| Kimi K3 | 17.4 | - | - |
| Gemini 4 Carbon | 59.4 | 50.4 | - |
| Claude Opus 5.5 | 58.4 | 51.7 | - |
| GPT-6 Astra | 56.9 | 49.1 | 96.1 |
| GPT-6.1 Sol | 54.5 | 46.8 | - |
| Claude Sonnet 5.5 | 54.2 | 50.5 | - |
Source: Artificial Analysis (2026-10-31).
Popularity
Kimi K3 has 99,214 downloads in the last month on HuggingFace and 8,653 likes.
Frequently asked
Quick answers to common questions
How much VRAM does Kimi K3 need?
Kimi K3 with 2779.9B parameters needs approximately 1612 GB at Q4_K_M quantization. Use our VRAM calculator for an exact estimate.
Is Kimi K3 better than other moonshotai models?
Kimi K3 has 2779.9B parameters with 1,048,576 context - a strong choice for general use.
What license is Kimi K3 under?
Kimi K3 is released under the other license, making it suitable for most commercial and personal projects.
What hardware runs Kimi K3 well?
With 2779.9B parameters, Kimi K3 requires adequate VRAM. High-end GPUs like the RTX 4090 (24GB), RTX 5090 (32GB), or Mac Studio with unified memory are good options. Check our hardware directory for specific recommendations.
What is the best quantization for Kimi K3?
Q4_K_M is the recommended sweet spot - ~98% of FP16 quality at ~27% of the size. Q5_K_M (~1974 GB) is an option if you have spare VRAM. Use our VRAM calculator to compare.
How long can Kimi K3's context window handle?
Kimi K3 supports a 1,048,576-token context window - enough for very long documents, codebases, or multi-turn conversations. Real-world usable context may vary by implementation.
What models compete with Kimi K3?
Kimi K3 competes with other 1390B–4170B. Browse our model directory for comparisons, benchmarks, and community reviews to find the best fit.
Nearby options
Similar models and compatible hardware by spec