--- title: "Kimi K3 | Aguasto" description: "Research reference for Kimi K3: official checkpoint, license and memory considerations. Documented in Aguasto Research; no Aguasto deployment or API availability is represented." canonical: "https://www.aguasto.com/models/kimi-k3" last-updated: "2026-10-13" --- # Kimi K3 | Aguasto Canonical HTML: [Kimi K3](https://www.aguasto.com/models/kimi-k3) > Research reference for Kimi K3: official checkpoint, license and memory considerations. Documented in Aguasto Research; no Aguasto deployment or API availability is represented. [Aguasto](https://www.aguasto.com/) [Model Index](https://www.aguasto.com/models) Kimi K3 [Resources](https://www.aguasto.com/resources) [GPU Library](https://www.aguasto.com/gpus) [Model Index](https://www.aguasto.com/models) [Inference Stack](https://www.aguasto.com/inference-stack) [Learn](https://www.aguasto.com/learn) [Insights](https://www.aguasto.com/insights) [Research](https://www.aguasto.com/research) [Engineering tools](https://www.aguasto.com/tools) Moonshot AI / Open-weight checkpoint reference ## Kimi K3 Documented in Aguasto Research. This checkpoint has not been evaluated or deployed by Aguasto and is not available through an Aguasto API. Original Aguasto conceptual illustration. Not manufacturer geometry, an operating Aguasto asset, or a measured benchmark. Aguasto Research documents third-party hardware, models and software. References do not represent Aguasto inventory, evaluated deployments or available inference services. ### Checkpoint identity Official identifier moonshotai/Kimi-K3[Moonshot AI KimiK3 official model card (external source)](https://huggingface.co/moonshotai/Kimi-K3) · [KimiK3 checkpoint configuration (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/config.json) · [KimiK3 version-specific license (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE) Architecture Multimodal mixture of experts with Kimi Delta Attention, gated MLA and Attention Residuals[Moonshot AI KimiK3 official model card (external source)](https://huggingface.co/moonshotai/Kimi-K3) · [KimiK3 checkpoint configuration (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/config.json) · [KimiK3 version-specific license (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE) Total parameters 2,800 billionPublished rounded counts; total stored weights determine the preliminary footprint, including all MoE experts.[Moonshot AI KimiK3 official model card (external source)](https://huggingface.co/moonshotai/Kimi-K3) · [KimiK3 checkpoint configuration (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/config.json) · [KimiK3 version-specific license (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE) Active parameters 104 billion per tokenActive parameters are a compute-related count, not the model's full stored weight count.[Moonshot AI KimiK3 official model card (external source)](https://huggingface.co/moonshotai/Kimi-K3) · [KimiK3 checkpoint configuration (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/config.json) · [KimiK3 version-specific license (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE) Modalities text-input, image-input, video-input, text-output[Moonshot AI KimiK3 official model card (external source)](https://huggingface.co/moonshotai/Kimi-K3) · [KimiK3 checkpoint configuration (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/config.json) · [KimiK3 version-specific license (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE) Context 1,048,576 tokensNative configuration where directly verified. Extensions, advertised K-labels and configuration differences are preserved in the notes below.[Moonshot AI KimiK3 official model card (external source)](https://huggingface.co/moonshotai/Kimi-K3) · [KimiK3 checkpoint configuration (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/config.json) · [KimiK3 version-specific license (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE) Release date Exact release date not independently established in the reviewed artifacts[Moonshot AI KimiK3 official model card (external source)](https://huggingface.co/moonshotai/Kimi-K3) · [KimiK3 checkpoint configuration (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/config.json) · [KimiK3 version-specific license (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE) Download access Public checkpoint reference[Moonshot AI KimiK3 official model card (external source)](https://huggingface.co/moonshotai/Kimi-K3) · [KimiK3 checkpoint configuration (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/config.json) · [KimiK3 version-specific license (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE) License Kimi K3 License (custom)This checkpoint uses a custom license, distinct from the older Kimi K2 Modified MIT terms. For a Model-as-a-Service operator, aggregate licensee-and-affiliate revenue above US$20 million over any consecutive 12 months requires a separate Moonshot agreement before commercial use. A separate user-interface attribution condition applies above 100 million MAU or US$20 million monthly revenue, with specified internal-use and official/certified-partner exceptions. Read the exact definitions and notice requirements.[Moonshot AI KimiK3 official model card (external source)](https://huggingface.co/moonshotai/Kimi-K3) · [KimiK3 checkpoint configuration (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/config.json) · [KimiK3 version-specific license (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE) ### Theoretical weight-memory footprint Using the publisher's approximately 2800 billion total parameters at 16 bits per parameter gives **5,600 GB** (5215.4 GiB) of nominal weights. weight GB = total parameters in billions × bits per weight / 8 This estimate excludes KV cache, activations, quantization metadata, non-weight buffers and serving overhead. It is not a tested GPU fit. MoE routing activates only part of the network per token while the complete expert weights still need storage. [Explore the memory estimator](https://www.aguasto.com/tools) #### Quantization The exact checkpoint config describes MXFP4 packed weight groups with four-bit storage and group size 32. That does not mean all tensors, activations or cache entries use four bits. Pin the compressed-tensors path and model revision for evaluation. [Moonshot AI KimiK3 official model card (external source)](https://huggingface.co/moonshotai/Kimi-K3) · [KimiK3 checkpoint configuration (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/config.json) · [KimiK3 version-specific license (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE) ### Serving considerations - The card's 2.8 trillion total and 104 billion activated parameter counts describe different boundaries. Expert-routing work does not reduce the weight inventory to 104B; the rounded total is the publisher's figure, not a counted Aguasto artifact. - The card and config agree on 1,048,576 context positions. Its attention-layer composition combines 69 KDA and 24 gated MLA layers, so a single conventional all-global attention-cache formula is insufficient for precise sizing. - The publisher recommends vLLM and SGLang and links model-specific recipes. Their version, custom model-code path, compressed-tensor support and exact hardware requirements must be pinned and tested before any compatibility or fit claim. - Text/image/video understanding adds preprocessing and visual-token costs. Record input resolutions, frame strategy, reasoning settings, concurrency and prompt/output limits in any future evaluation. - Aguasto has not hosted, benchmarked or commercially cleared this checkpoint. This page documents technical evaluation considerations and prospective license review, not an available model service. [Moonshot AI KimiK3 official model card (external source)](https://huggingface.co/moonshotai/Kimi-K3) · [KimiK3 checkpoint configuration (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/config.json) · [KimiK3 version-specific license (external source)](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE) #### Framework documentation These are third-party documentation starting points. Support depends on the exact checkpoint, runtime version and hardware. ### Continue the research [vLLM](https://www.aguasto.com/inference-stack/vllm) [SGLang](https://www.aguasto.com/inference-stack/sglang) [Compare model references](https://www.aguasto.com/models/compare) [GPU specification library](https://www.aguasto.com/gpus) [Weights, cache and serving overhead](https://www.aguasto.com/learn/model-weights-and-runtime-memory) ### Sources & verification The register preserves each source's publication and verification date. Manufacturer claims and documentation are attributed; Aguasto has not run the cited products in production. - [Moonshot AI KimiK3 official model card ↗](https://huggingface.co/moonshotai/Kimi-K3) Exact checkpoint identity and open weight listing; Publisher-rounded2.8Ttotal104Bactivated parameters; 1048576 context positions; 69KDA24gatedMLAattention composition and multimodal inputs; Explicit vLLM/SGLang model-specific recipe links. Publication: date not stated in the reviewed source. Verified: 2026-10-13. - [KimiK3 checkpoint configuration ↗](https://huggingface.co/moonshotai/Kimi-K3/blob/main/config.json) 1048576 max\_position\_embeddings; Kimi attention configuration; MXFP4 packed quantized groups, group size32 and4bits. Publication: date not stated in the reviewed source. Verified: 2026-10-13. - [KimiK3 version-specific license ↗](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE) Custom license and notice duties; MaaS aggregate revenue\>US$20m over consecutive12months separate-agreement condition; Separate\>100mMAU or\>US$20mmonthly-revenue attribution condition; Specified internal-use and official/certified-partner exceptions. Publication: date not stated in the reviewed source. Verified: 2026-10-13. ### Reference journey [Compare models ↗](https://www.aguasto.com/models/compare) [Estimate weights ↗](https://www.aguasto.com/tools) [Hardware references ↗](https://www.aguasto.com/gpus) [Aguasto inference strategy ↗](https://www.aguasto.com/ai-inference) [Sources & verification ↓](https://www.aguasto.com/models/kimi-k3#sources) Research reference · No available Aguasto service ## Company navigation - [Home](https://www.aguasto.com/) - [Infrastructure](https://www.aguasto.com/infrastructure) - [GPU Cloud](https://www.aguasto.com/gpu-cloud) - [AI Inference](https://www.aguasto.com/ai-inference) - [Development & Investment](https://www.aguasto.com/development) - [Company](https://www.aguasto.com/company) - [Contact](https://www.aguasto.com/contact) - [Privacy](https://www.aguasto.com/privacy) - [Resources](https://www.aguasto.com/resources) - [GPU Library](https://www.aguasto.com/gpus) - [Model Index](https://www.aguasto.com/models) - [Inference Stack](https://www.aguasto.com/inference-stack) - [Learn](https://www.aguasto.com/learn) - [Insights](https://www.aguasto.com/insights) - [Research](https://www.aguasto.com/research) - [Industry cases](https://www.aguasto.com/research/cases) - [Why Modular](https://www.aguasto.com/why-modular) - [Partners](https://www.aguasto.com/partners) - [GPU comparison](https://www.aguasto.com/gpus/compare) - [Model comparison](https://www.aguasto.com/models/compare) - [Engineering tools](https://www.aguasto.com/tools) - [Development friction](https://www.aguasto.com/research/hyperscale-development-friction) - [NVIDIA H100 SXM](https://www.aguasto.com/gpus/h100) - [NVIDIA H200 SXM](https://www.aguasto.com/gpus/h200) - [NVIDIA B200 SXM](https://www.aguasto.com/gpus/b200) - [NVIDIA B300](https://www.aguasto.com/gpus/b300) - [NVIDIA GB200 NVL72](https://www.aguasto.com/gpus/gb200-nvl72) - [NVIDIA GB300 NVL72](https://www.aguasto.com/gpus/gb300-nvl72) - [NVIDIA Vera Rubin NVL72](https://www.aguasto.com/gpus/vera-rubin-nvl72) - [AMD Instinct MI300X](https://www.aguasto.com/gpus/mi300x) - [AMD Instinct MI325X](https://www.aguasto.com/gpus/mi325x) - [AMD Instinct MI355X](https://www.aguasto.com/gpus/mi355x) - [AMD Instinct MI455X](https://www.aguasto.com/gpus/mi455x) - [AMD Helios](https://www.aguasto.com/gpus/helios) - [GLM-4.5](https://www.aguasto.com/models/glm-4-5) - [Qwen3-235B-A22B](https://www.aguasto.com/models/qwen3-235b-a22b) - [DeepSeek-R1](https://www.aguasto.com/models/deepseek-r1) - [Llama 3.3 70B Instruct](https://www.aguasto.com/models/llama-3-3-70b-instruct) - [Gemma 3 27B IT](https://www.aguasto.com/models/gemma-3-27b-it) - [gpt-oss-120b](https://www.aguasto.com/models/gpt-oss-120b) - [Kimi-K2-Instruct](https://www.aguasto.com/models/kimi-k2-instruct) - [GLM-5.3](https://www.aguasto.com/models/glm-5-3) - [Qwen3.8-27B](https://www.aguasto.com/models/qwen3-8-27b) - [DeepSeek-V4.1-Flash](https://www.aguasto.com/models/deepseek-v4-1-flash) - [Llama 4 Scout 17B-16E Instruct](https://www.aguasto.com/models/llama-4-scout-17b-16e-instruct) - [Gemma 4 31B IT](https://www.aguasto.com/models/gemma-4-31b-it) - [Kimi K3](https://www.aguasto.com/models/kimi-k3) - [vLLM](https://www.aguasto.com/inference-stack/vllm) - [SGLang](https://www.aguasto.com/inference-stack/sglang) - [TensorRT-LLM](https://www.aguasto.com/inference-stack/tensorrt-llm) - [NVIDIA Dynamo](https://www.aguasto.com/inference-stack/dynamo) - [NVIDIA Triton Inference Server](https://www.aguasto.com/inference-stack/triton) - [AMD ROCm](https://www.aguasto.com/inference-stack/rocm) - [Ray Serve](https://www.aguasto.com/inference-stack/ray-serve) - [Kubernetes GPU orchestration — NVIDIA device plugin](https://www.aguasto.com/inference-stack/kubernetes) - [Model weights are only the beginning of the memory budget](https://www.aguasto.com/learn/model-weights-and-runtime-memory) - [Why tokens per second is not enough to benchmark inference](https://www.aguasto.com/research/inference-benchmark-methodology) - [Prefill, decode and the cost of separating them](https://www.aguasto.com/learn/prefill-decode-and-disaggregated-serving) - [Choosing between vLLM, SGLang and TensorRT-LLM](https://www.aguasto.com/learn/choosing-an-inference-serving-framework) - [From Power to Inference—Building a Distributed AI Compute Network](https://www.aguasto.com/learn/from-power-to-inference) - [Dedicated compute and shared inference: compare the whole workload](https://www.aguasto.com/insights/dedicated-and-shared-inference-economics) - [Distributed inference reliability begins with the failure boundary](https://www.aguasto.com/learn/distributed-inference-reliability) - [Core Scientific: reading contracted, energized and billed capacity](https://www.aguasto.com/research/cases/core-scientific-capacity-disclosures) - [Why Modular AI Infrastructure Matters Now](https://www.aguasto.com/research/why-modular-ai-infrastructure-matters-now) - [What Is a Modular AI Data Center?](https://www.aguasto.com/learn/what-is-a-modular-ai-data-center) - [How Modular AI Data Centers Are Built and Commissioned](https://www.aguasto.com/learn/modular-data-center-factory-to-commissioning) - [Liquid Cooling for High-Density GPU Systems: Follow the Interfaces](https://www.aguasto.com/learn/liquid-cooling-gpu-rack-interfaces) - [Why Electricity Availability Shapes AI Infrastructure Deployment](https://www.aguasto.com/research/electricity-availability-and-time-to-power) - [What Is KV Cache? Capacity, Reuse and Isolation](https://www.aguasto.com/learn/kv-cache-capacity-and-policy) - [Continuous Batching: Throughput Without Hiding Latency](https://www.aguasto.com/learn/continuous-batching-and-latency) - [Applied Digital: Reading the Polaris Forge 1 Delivery Milestones](https://www.aguasto.com/research/cases/applied-digital-polaris-forge-capacity-milestones) - [Crusoe Abilene: Separate Delivered Infrastructure from Expansion Plans](https://www.aguasto.com/research/cases/crusoe-abilene-delivery-and-expansion) - [Official LinkedIn](https://www.linkedin.com/company/aguasto) This is a generated representation of the public HTML page. Diagrams are conceptual. GPU and inference services are planned and are not currently available to book or call. The contact page describes the available inquiry flow.