- Size
- 1.0T parameterscounted from the safetensors header on the Hub
- Licence
- unknown, MiniMax Model LicenseOur reading, check the licence: Despite the id, this text is the MiniMax Model License (dated for MiniMax-M2.5), not a modified MIT licence: it lets you use, copy, change and distribute the model with no limit on commercial use and no user or revenue threshold. If you distribute the model or a product or service that uses it, you must include the licence, mark changed files and keep a Notice file, though you may license your own changes differently. A Prohibited Uses Policy is part of the licence and bars, among other things, any military purpose and fully automated decisions that adversely affect someone's legal rights.
- On the Hub since
- 11 Jul 2025the date the repo was created
- Context length
- 128k tokensfrom the model's config
- Kind
- Base model (no base model declared on the Hub)DeepseekV3ForCausalLM
- On Hugging Face
- moonshotai/Kimi-K2-Instruct232k downloads in the last 30 days, 3.0k likes
Scores
- Edit format: diff59.1%Aider LLM Leaderboards
Epoch's name: Kimi-K2-Instruct. Best closed model on this set: gpt-5-2025-08-07_high, 88.0% (OpenAI). All 72 results
- default setting61.1%Fiction.live leaderboard
Epoch's name: Kimi-K2-Instruct. Best closed model on this set: gpt-5-2025-08-07_medium, 97.2% (OpenAI). All 61 results
- Scaffold: OpenHands4.9%26 Aug 2025, gso-bench.github.io
Epoch's name: Kimi-K2-Instruct. Best closed model on this set: claude-fable-5-1_unknown, 88.2% (Anthropic). All 42 results
- default setting85.6%lechmazur/writing Github repository
Epoch's name: Kimi-K2-Instruct. Best closed model on this set: gpt-5-2025-08-07_medium, 86.0% (OpenAI). All 49 results
- default setting26.3%simple-bench.com
Epoch's name: Kimi-K2-Instruct. Best closed model on this set: claude-fable-5-2_unknown, 94.9% (Anthropic). All 125 results
- Agent: Terminus 227.8%±2.51 Nov 2025, Terminal-Bench v2 Leaderboard
- Agent: Terminus 227.8%±2.511 Jul 2025, tbench.ai
- Agent: OpenHands26.7%±2.711 Jul 2025, tbench.ai
- Agent: OpenHands26.7%±2.72 Nov 2025, Terminal-Bench v2 Leaderboard
Epoch's name: Kimi-K2-Instruct. Best closed model on this set: gpt-5.5_unknown, 84.7% (OpenAI). All 202 results
- default setting39.4%WeirdML Leaderboard
Epoch's name: Kimi-K2-Instruct. Best closed model on this set: gpt-6-astra_promax, 93.6% (OpenAI). All 173 results
Claimed by the maker Claimed
The model card has no claimed scores in a machine-readable form (the card's model-index), so none are shown. The maker may give scores in the card text: read the card.
Other views, which we link to and do not copy: look this model up on Artificial Analysis and the Arena leaderboard.
Independent scores from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Each row links its own source. What the badges mean.
Run it yourself
Memory needed = weights + context cache + 0.50 GB for the runtime. Context cache at 8k tokens: 0.54 GB, worked out from the model's config (moonshotai/Kimi-K2-Instruct), at 16-bit precision. Choose a machine above to see a verdict for each format. Method
GGUF for llama.cpp, Ollama, LM Studio and others
unsloth/Kimi-K2-Instruct-GGUF (converted by unsloth), 20k downloads. Sizes are the real files on the Hub, all parts added up.
| Format | Size | Memory needed | Get it |
|---|---|---|---|
| UD-IQ3_XXS | 388.0 GBfile size, 9 parts | 389.0 GB388.0 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:UD-IQ3_XXSDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "UD-IQ3_XXS/Kimi-K2-Instruct-UD-IQ3_XXS-*-of-00009.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| Q3_K_S | 412.0 GBfile size, 10 parts | 413.1 GB412.0 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:Q3_K_SDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "Q3_K_S/Kimi-K2-Instruct-Q3_K_S-*-of-00010.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| UD-Q3_K_XL | 421.0 GBfile size, 10 parts | 422.1 GB421.0 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:UD-Q3_K_XLDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "UD-Q3_K_XL/Kimi-K2-Instruct-UD-Q3_K_XL-*-of-00010.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| Q3_K_M | 455.8 GBfile size, 11 parts | 456.8 GB455.8 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:Q3_K_MDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "Q3_K_M/Kimi-K2-Instruct-Q3_K_M-*-of-00011.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| IQ4_XS | 509.0 GBfile size, 12 parts | 510.0 GB509.0 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:IQ4_XSDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "IQ4_XS/Kimi-K2-Instruct-IQ4_XS-*-of-00012.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| IQ4_NL | 538.8 GBfile size, 12 parts | 539.8 GB538.8 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:IQ4_NLDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "IQ4_NL/Kimi-K2-Instruct-IQ4_NL-*-of-00012.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| Q4_0 | 540.7 GBfile size, 13 parts | 541.8 GB540.7 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:Q4_0Download hf download unsloth/Kimi-K2-Instruct-GGUF --include "Q4_0/Kimi-K2-Instruct-Q4_0-*-of-00013.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| Q4_K_S | 542.7 GBfile size, 13 parts | 543.8 GB542.7 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:Q4_K_SDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "Q4_K_S/Kimi-K2-Instruct-Q4_K_S-*-of-00013.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| UD-Q4_K_XL | 546.8 GBfile size, 13 parts | 547.8 GB546.8 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:UD-Q4_K_XLDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "UD-Q4_K_XL/Kimi-K2-Instruct-UD-Q4_K_XL-*-of-00013.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| Q4_K_M | 578.1 GBfile size, 13 parts | 579.2 GB578.1 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:Q4_K_MDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "Q4_K_M/Kimi-K2-Instruct-Q4_K_M-*-of-00013.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| Q4_1 | 598.4 GBfile size, 13 parts | 599.4 GB598.4 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:Q4_1Download hf download unsloth/Kimi-K2-Instruct-GGUF --include "Q4_1/Kimi-K2-Instruct-Q4_1-*-of-00013.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| Q5_K_S | 658.0 GBfile size, 15 parts | 659.1 GB658.0 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:Q5_K_SDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "Q5_K_S/Kimi-K2-Instruct-Q5_K_S-*-of-00015.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| Q5_K_M | 678.3 GBfile size, 16 parts | 679.4 GB678.3 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:Q5_K_MDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "Q5_K_M/Kimi-K2-Instruct-Q5_K_M-*-of-00016.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| UD-Q5_K_XL | 680.4 GBfile size, 16 parts | 681.4 GB680.4 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:UD-Q5_K_XLDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "UD-Q5_K_XL/Kimi-K2-Instruct-UD-Q5_K_XL-*-of-00016.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| Q6_K | 784.8 GBfile size, 18 parts | 785.8 GB784.8 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:Q6_KDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "Q6_K/Kimi-K2-Instruct-Q6_K-*-of-00018.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| UD-Q6_K_XL | 818.7 GBfile size, 19 parts | 819.7 GB818.7 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:UD-Q6_K_XLDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "UD-Q6_K_XL/Kimi-K2-Instruct-UD-Q6_K_XL-*-of-00019.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| Q8_0 | 1016.1 GBfile size, 23 parts | 1017.2 GB1016.1 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:Q8_0Download hf download unsloth/Kimi-K2-Instruct-GGUF --include "Q8_0/Kimi-K2-Instruct-Q8_0-*-of-00023.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| UD-Q8_K_XL | 1108.3 GBfile size, 25 parts | 1109.4 GB1108.3 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:UD-Q8_K_XLDownload hf download unsloth/Kimi-K2-Instruct-GGUF --include "UD-Q8_K_XL/Kimi-K2-Instruct-UD-Q8_K_XL-*-of-00025.gguf" --local-dir Kimi-K2-Instruct-GGUF |
| BF16 | 1912.2 GBfile size, 45 parts | 1913.2 GB1912.2 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Kimi-K2-Instruct-GGUF:BF16Download hf download unsloth/Kimi-K2-Instruct-GGUF --include "BF16/Kimi-K2-Instruct-BF16-*-of-00045.gguf" --local-dir Kimi-K2-Instruct-GGUF |
MLX for Apple silicon
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| MLX 3bitmlx-community/moonshotai_Kimi-K2-Instruct-mlx-3bit | 418.5 GBfile size | 419.5 GB418.5 + 0.54 + 0.50 | Download page |
| MLX 4bitmlx-community/Kimi-K2-Instruct-4bit | 537.9 GBfile size | 539.0 GB537.9 + 0.54 + 0.50 | Download page |
Original weights size from the safetensors header, for vLLM, SGLang, Transformers
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| F8_E4M3original weights | 958.3 GBfile size | 959.3 GB958.3 + 0.54 + 0.50 |
hf download moonshotai/Kimi-K2-Instruct --local-dir Kimi-K2-InstructOther conversions (4; sizes not read)
- unsloth/Kimi-K2-Instruct FP8, 210 downloads
- luisra/Kimi-K2-Instruct-4bit bitsandbytes, 4bit, 175 downloads
- RedHatAI/Kimi-K2-Instruct-quantized.w4a16 GPTQ, w4a16, 131 downloads
- OPEA/Kimi-K2-Instruct-int4-mixed-AutoRound Other, int4, 61 downloads
The Hub tags 18 quantised repos as conversions of this model; we read the most downloaded and show the maker's own first.
Run it by API
2 providers serve this model through the Hugging Face router. Prices are US dollars per million tokens. Precision is what the provider reports; a lower precision than the original weights can change answers. Live first, then cheapest output.
| Provider | In | Out | Context | Precision | Tokens/s | First token | Uptime 24h | Tools | JSON |
|---|---|---|---|---|---|---|---|---|---|
| Novita | $0.57 | $2.30 | 128k | fp8 | 32 | 2.5 s | 100.0% | yes | no |
| featherless-ai | ? | ? | ? | not given | ? | ? | ? | not given | not given |
Through the router, with a Hugging Face token (the router bills your Hugging Face account). Change the name after the colon to pick another provider.
curl https://router.huggingface.co/v1/chat/completions \
-H "Authorization: Bearer $HF_TOKEN" -H "Content-Type: application/json" \
-d '{"model": "moonshotai/Kimi-K2-Instruct:novita", "messages": [{"role": "user", "content": "Hello"}]}'Not built yet: one key for every open model, with BlockRun, and our own tested-quality score per provider. Until then, precision and measured speed are the best guide to a provider here.
Planned for this page Not built yet
- Try it in private chat
- A link to Solana AI's confidential chat for the models it serves. This model is not listed as served there.
- Fine-tune it
- Fine-tuning a few base models from this page and publishing the result here.
- Creators and earnings
- Pages for people who publish fine-tunes and adapters of this model, and what they earn.
- Fund and bounties
- Bounties for new open models, paid from a public fund. How the fund will work.
- Contribute compute
- How to give compute to open model work, with Akash.
Model data from the Hugging Face Hub 19 Nov 2026, 13:43 UTC. The repo. Found a mistake on this page? Tell us.