Kimi-K2-Instruct-0905
Kimi-K2-Instruct-0905 is an open-weight language model from Moonshot AI with 1.03T parameters (mixture of experts). At 4-bit it needs about 693.2 GB of memory, more than any single device in our directory has.
Overview
Includes room for an 8k context and runtime overhead. More context needs more memory.
Key information
- Lab
- Moonshot AI
- Released
- 3 Sep 2025
- Parameters
- 1.03T · MoE
- Type
- Language
- Context
- 262,144 tokens
- License
- Modified MIT
- HF downloads, 30 days
- 8,724
- Repository
- moonshotai/Kimi-K2-Instruct-0905
More from Moonshot AI
View allWill it run on your hardware?
How we estimate| Estimate or device | Memory |
|---|---|
| Q4 | 693.2 GB |
| Q8 | 1212.6 GB |
| BF16 | 2273.9 GB |
| RTX 5060 Ti 16 GB | 16 GB usable |
| RTX 5090 32 GB | 32 GB usable |
| DGX Spark 128 GB | 108.8 GB usable |
| Mac Studio M5 Ultra 256 GB | 192 GB usable |
Single-device memory estimate. Pick a device to update the On your device column.
36 GB usable memory
| Format | Memory needed | On your device | Example hardware |
|---|---|---|---|
| 4-bit (Q4) | ≈ 693.2 GB | Too big 1926% | No single device in our directory has enough memory |
| 8-bit (Q8) | ≈ 1212.6 GB | Too big 3368% | No single device in our directory has enough memory |
| 16-bit (BF16) | ≈ 2273.9 GB | Too big 6316% | No single device in our directory has enough memory |
Fits means the estimate stays within 90% of usable memory. Tight is above 90% and up to 100%. Too big is above 100%. These are memory checks, not speed or compatibility tests.
Mac allowances are 66% of total RAM up to 36 GB, and 75% above. AI boxes allow 85% because their memory is shared with the operating system.
Where to run it
On a Mac
Start with the MLX files, built for Apple Silicon.
mlx-community/Kimi-K2-Instruct-0905-mlx-3bitCheck your MacWith GGUF
Choose a file for llama.cpp or a compatible app.
unsloth/Kimi-K2-Instruct-0905-GGUFFind compatible hardwareThrough an API
Novita (fp8) is the cheapest provider we found.
$0.60 / $2.50
Input / output per 1M tokens
₹57 / ₹236 at ₹94.29 per $1
Blended: $1.075 (₹101) per 1M. Ranked by a 3:1 mix of input and output tokens.
mlx_lm.generate --model mlx-community/Kimi-K2-Instruct-0905-mlx-3bit \
--prompt "Explain how a transformer works."llama-cli -hf unsloth/Kimi-K2-Instruct-0905-GGUF \
-c 8192 -p "Explain how a transformer works."Example commands built from the repository ids above. They need a working install of MLX or llama.cpp. We have not run them; check each repository for available files.
| Precision | |||||
|---|---|---|---|---|---|
| NovitaCheapest | $1.075₹101 | $0.60₹57 | $2.50₹236 | 262,144 | FP8 |
INR at ₹94.29 per $1. Prices fetched 14 Nov 2026.
More about this model
Frequently asked questions
How much memory does Kimi-K2-Instruct-0905 need? At an 8k context, about 693.2 GB at 4-bit, 1212.6 GB at 8-bit and 2273.9 GB at 16-bit (BF16). A longer context needs more.
Can I run Kimi-K2-Instruct-0905 on a Mac? No Mac in our directory has enough usable memory for the 4-bit estimate of 693.2 GB. An MLX build is available: mlx-community/Kimi-K2-Instruct-0905-mlx-3bit.
Does the full context fit? Kimi-K2-Instruct-0905 supports up to 262,144 tokens of context. Our memory estimates use 8k, so long prompts need more memory than shown.
What does it cost through an API? The cheapest provider we found is Novita, at $0.60 per 1M input tokens and $2.50 per 1M output tokens (₹57 and ₹236 at 94.29 rupees per dollar). Ranked by a 3:1 mix of input and output tokens.
License in plain English
Modified MIT. This model adds conditions to the MIT license. The conditions differ by model; read its own license before commercial use. This summary is not legal advice.
OMHQ RigBench
Not yet tested. We show no speed or quality score without a measurement.
Revision history and sources
Last updated . Sources:Hugging Face model card, MLX repository, GGUF repository, OpenRouter.