July 27, 2026
Moonshot AI published the weights of Kimi K3 on Hugging Face on July 27, according to the repository’s commit history. Hugging Face counts 2,779,931,837,184 parameters in the safetensors files, matching the 2.8 trillion on the model card, and the weight files total about 1,561 GB.
Takeaway points
- The model’s config lists 896 experts, of which 16 are chosen per token alongside 2 shared experts, and a maximum of 1,048,576 positions.
- The Kimi K3 License requires a separate agreement with Moonshot before commercial use by any model-as-a-service business with more than $20 million in revenue over 12 months.
- Moonshot’s own benchmark table puts Kimi K3 behind GPT-5.6 Sol on DeepSWE coding, 67.5 against 73.0, and behind Claude Fable 5 on OSWorld 2.0.
Kimi K3 architecture and parameters, from the model card
Moonshot calls Kimi K3 “the world’s first open 3T-class model” and its most capable model to date. It is a mixture-of-experts model with 104 billion parameters active per token, built on two techniques the company calls Kimi Delta Attention and Attention Residuals. Of its 93 layers, 69 use Kimi Delta Attention and 24 use gated multi-head latent attention, and one is dense. The card lists 96 attention heads, a latent MoE dimension of 3,584, an expert hidden size of 3,072, a vocabulary of 160K tokens and a vision encoder, MoonViT-V2, with 401 million parameters.
Moonshot said its Stable LatentMoE framework, which activates 16 of the 896 experts, gives “an approximate 2.5” times improvement in scaling efficiency over Kimi K2. The weights are stored in MXFP4 with MXFP8 activations, which the card says came from quantization-aware training rather than compression after the fact.
The card is inconsistent on one point. Its introduction says the model “understands text, images, and video within the same model,” while its summary table lists the modality as “Text, Image.” Hugging Face files Kimi K3 as an image-text-to-text model. The repository holds 115 files at its root, and the config names the architecture KimiK3ForConditionalGeneration with a hidden size of 7,168.
Kimi K3 benchmarks against GPT-5.6 Sol and Claude Fable 5
Moonshot reports every Kimi K3 result with reasoning effort set to max and a temperature of 1.0. Against GPT-5.6 Sol, Claude Fable 5, Claude Opus 4.8, GPT-5.5 and GLM-5.2, the picture is mixed. Kimi K3 leads on BrowseComp, 91.2 against 90.4 for GPT-5.6 Sol, and on SWE-Marathon, a long-running coding test, with 42.0 against 39.0 for GPT-5.6 Sol and 35.0 for Fable 5. It scores 94.5 on MCPMark-Verified, against 92.9 for GPT-5.6 Sol and 87.4 for Fable 5. Its 74.7 on AA-LCR, a long-context reasoning test, is the highest in the table.
It trails elsewhere. On GPQA Diamond it scores 93.5 against 94.1 for GPT-5.6 Sol. On HLE-Full it scores 43.5 without tools and 56.0 with them, against 53.3 and 63.0 for Fable 5, and on Agents’ Last Exam it scores 28.3, between Fable 5 at 25.7 and GPT-5.6 Sol at 29.6. On Terminal-Bench 2.1 it scores 88.3 against 88.8. On OSWorld 2.0, a computer-use test, it scores 58.3 against 66.1 for Fable 5 and 62.6 for GPT-5.6 Sol. On FrontierSWE it scores 81.2, behind Fable 5 at 86.6 but ahead of GPT-5.6 Sol at 71.3, and on GDPval-AA, an Elo rating of professional work, it scores 1,686 against 1,747 for Fable 5 and 1,736 for GPT-5.6 Sol.
The table covers 45 benchmarks across reasoning, coding, agentic work and vision. Vision is where Kimi K3 does best against the field: it leads on OmniDocBench with 91.1 and on Video-MME with 90.0. Two of its reasoning scores, CritPt and AA-LCR, are cited from Artificial Analysis as of July 23 rather than run by Moonshot.
The harness matters to these numbers. Moonshot ran Kimi K3 on DeepSWE with its own Kimi Code harness; on the official leaderboard’s mini-SWE-agent harness, the card says, it scores 67.3. Its SWE-Marathon runs used a version of the tasks recalibrated for Nvidia H20 hardware, taken before the benchmark’s final v1.1 release. Community evaluation results for four benchmarks were added to the repository on the day of release, and a Toolathlon-Verified result followed on September 2. The same afternoon as the first commit, a Hugging Face staff change to enable inference providers was merged into the repository.
Kimi K3 license terms for commercial use
The licence opens with a broad grant in the style of the MIT licence, allowing anyone to use, copy, modify, fine-tune and sell the software and weights. Two conditions follow. A business that offers model inference or fine-tuning as a service, with more than $20 million in revenue over any consecutive 12 months, must sign a separate agreement with Moonshot before commercial use. Any commercial product built on the model with more than 100 million monthly active users, or more than $20 million in monthly revenue, must display “Kimi K3” prominently.
Neither condition applies to internal use, which the licence defines as use that does not make the software, its outputs or its capabilities available to third parties, or to use through Moonshot’s own products and certified inference partners. Hugging Face labels the licence “other.”
Kimi K3 download size and hardware requirements
The official weights come to about 1,561 GB, and the card recommends the vLLM, SGLang and TokenSpeed inference engines. StrideNote’s own sum of the published GGUF files found that the smallest Kimi K3 build is 589 GB, more than twelve times the 48 GB of unified memory on StrideNote’s test machine. How GGUF builds of one model differ explains why a filename alone does not show what a build contains.
Demand has been steady. By September 10 the repository showed 2,319,885 downloads and 11,269 likes.
What Moonshot has not published about Kimi K3
The model card does not state how much data Kimi K3 was trained on or on what hardware. It links to a separate full report, which StrideNote has not reviewed. The card also does not reconcile its two statements about video input.
Sources: Kimi K3 model card; Kimi K3 License; config; commit history.