Hugging FaceStrong at coding and multi-step work for a 16GB Mac, in words only.
- Rating, estimated
- 3.8
- Downloads
- 41K
- Parameters
- 9B
- Context
- 512K
- Download
- 3.5 to 18 GB
- Commercial use
- Yes
Runs great on M4 · 16 GB
Outmake picks the Q4_K_M size: a 5.8 GB download that uses 6.4 GB of memory, and writes about 6.5 words (8.7 tokens) a second.
Good at
Licence
Apache 2.0. Commercial use: yes. The maker's page has the full terms.
K2-Horizon 7B runs well from a Mac with an M1 and 8 GB of memory, at Q3_K_M, writing about 4.2 words (5.6 tokens) a second. 59 of 59 Mac configurations run it well.
Compare with
The nearest models in size, judged for the same Mac.
- Qwen3.5 9BChat that reads imagesRuns great · 6.6 GB · 13 words/sGet
- Gemma 4 E4BUnderstands photos and speechRuns great · 7.6 GB · 8.9 words/sGet
- Gemma 4 12BChat with pictures and audioRuns great · 5.9 GB · 10 words/sGet
- K2-Horizon 3.7BFor tasks that need tools called along the way, with reasoning worked through before each answer.Runs great · 4.4 GB · 7.4 words/sGet
Sizes
| Size | Download | Memory at 8K | Memory at 32K | Open |
|---|---|---|---|---|
| Q3_K_MSmaller | 4.4 GB | 5.0 GB | 7.0 GB | Open |
| Q4_K_MBalanced | 5.8 GB | 6.4 GB | 8.3 GB | Open |
| Q6_KBest | 7.8 GB | 8.2 GB | 10 GB | Open |
All 22 sizes
| Size | Download | Memory at 8K | Memory at 32K | Open |
|---|---|---|---|---|
| IQ2_M | 3.5 GB | 4.1 GBestimated | 6.0 GB | Open |
| Q2_K | 3.6 GB | 4.2 GBestimated | 6.2 GB | Open |
| IQ3_XXS | 4.1 GB | 4.7 GBestimated | 6.6 GB | Open |
| Q3_K_S | 4.3 GB | 4.9 GBestimated | 6.8 GB | Open |
| IQ3_XS | 4.3 GB | 4.9 GBestimated | 6.8 GB | Open |
| Q3_K_L | 4.6 GB | 5.2 GBestimated | 7.2 GB | Open |
| IQ3_M | 4.9 GB | 5.5 GBestimated | 7.4 GB | Open |
| IQ4_XS | 5.2 GB | 5.9 GBestimated | 7.8 GB | Open |
| Q4_0 | 5.4 GB | 6.0 GBestimated | 8.0 GB | Open |
| Q4_K_S | 5.5 GB | 6.1 GBestimated | 8.1 GB | Open |
| IQ4_NL | 5.7 GB | 6.4 GBestimated | 8.3 GB | Open |
| Q4_1 | 5.9 GB | 6.5 GBestimated | 8.5 GB | Open |
| Q4_K_L | 6.2 GB | 6.8 GBestimated | 8.7 GB | Open |
| Q5_K_S | 6.5 GB | 7.1 GBestimated | 9.1 GB | Open |
| Q5_K_M | 6.9 GB | 7.2 GBestimated | 9.2 GB | Open |
| Q6_K_S | 7.5 GB | 7.9 GBestimated | 9.8 GB | Open |
| Q6_K_L | 8.1 GB | 8.5 GBestimated | 10 GB | Open |
| Q8_0 | 9.6 GB | 9.9 GBestimated | 12 GB | Open |
| BF16 | 18 GB | 18 GBestimated | 20 GB | Open |
Memory is what llama.cpp's own calculator reported for each size when the catalog was built, with the app's settings. Smaller, Balanced and Best are the three sizes the store names.
Run It Anywhere
K2-Horizon 7B is an open model, so any app that runs GGUF files can run it. These take the Q4_K_M size from bartowski straight from Hugging Face.
- LM StudioMac, Windows, Linux
Open - JanMac, Windows, Linux
Open - MstyMac, Windows, Linux
Open - OllamaMac, Windows, Linux
ollama run hf.co/bartowski/K2-Horizon-7B-GGUF:Q4_K_M - llama.cppMac, Windows, Linux
llama-server -hf bartowski/K2-Horizon-7B-GGUF:Q4_K_M - Docker Model RunnerMac, Windows, Linux
docker model run hf.co/bartowski/K2-Horizon-7B-GGUF:Q4_K_M
On a PC Graphics Card
| Video memory | Largest size that fits | Download | Memory at 8K |
|---|---|---|---|
| 6 GB | IQ4_XS | 5.2 GB | 5.9 GB |
| 8 GB | Q6_K_S | 7.5 GB | 7.9 GB |
| 10 GB | Q6_K_L | 8.1 GB | 8.5 GB |
| 11 GB | Q8_0 | 9.6 GB | 9.9 GB |
| 12 GB | Q8_0 | 9.6 GB | 9.9 GB |
| 16 GB | Q8_0 | 9.6 GB | 9.9 GB |
| 20 GB | BF16 | 18 GB | 18 GB |
| 24 GB | BF16 | 18 GB | 18 GB |
| 32 GB | BF16 | 18 GB | 18 GB |
| 48 GB | BF16 | 18 GB | 18 GB |
The largest size, of those that keep three bits a weight or more, whose memory at an 8K chat is under 92% of the card's; the rest is the desktop's. Estimated from the files, the same way as for a Mac; speed on a PC is not estimated yet. A size that doesn't fit the card still runs, slower, with part of it in ordinary memory.
On every Mac
| Mac | Size | Fit | Memory | Writes | Reads |
|---|---|---|---|---|---|
| M6 · 16 GB | Q4_K_M | Runs great | 6.4 GB | 9.2 | 699 |
| M6 · 24 GB | Q6_K | Runs great | 10 GB | 8 | 699 |
| M6 · 32 GB | Q6_K | Runs great | 10 GB | 8 | 699 |
| M5 · 16 GB | Q4_K_M | Runs great | 6.4 GB | 8.4 | 258 |
| M5 · 24 GB | Q6_K | Runs great | 10 GB | 7.2 | 258 |
| M5 · 32 GB | Q6_K | Runs great | 10 GB | 7.2 | 258 |
| M5 Pro · 24 GB | Q6_K | Runs great | 10 GB | 14 | 516 |
| M5 Pro · 48 GB | Q6_K | Runs great | 18 GB | 14 | 516 |
| M5 Pro · 64 GB | Q6_K | Runs great | 29 GB | 14 | 645 |
| M5 Max · 36 GB | Q6_K | Runs great | 10 GB | 21 | 1032 |
| M5 Max · 48 GB | Q6_K | Runs great | 18 GB | 28 | 1290 |
| M5 Max · 64 GB | Q6_K | Runs great | 29 GB | 28 | 1290 |
| M5 Max · 128 GB | Q6_K | Runs great | 29 GB | 28 | 1290 |
| M5 Ultra · 96 GB | Q6_K | Runs great | 29 GB | 52 | 2064 |
| M5 Ultra · 256 GB | Q6_K | Runs great | 29 GB | 52 | 2580 |
| M5 Ultra · 512 GB | Q6_K | Runs great | 29 GB | 52 | 2580 |
| M4 · 16 GB | Q4_K_M | Runs great | 6.4 GB | 6.5 | 87 |
| M4 · 24 GB | Q6_K | Runs great | 10 GB | 5.5 | 87 |
| M4 · 32 GB | Q6_K | Runs great | 10 GB | 5.5 | 87 |
| M4 Pro · 24 GB | Q6_K | Runs great | 10 GB | 12 | 173 |
| M4 Pro · 48 GB | Q6_K | Runs great | 18 GB | 12 | 173 |
| M4 Pro · 64 GB | Q6_K | Runs great | 29 GB | 12 | 173 |
| M4 Max · 36 GB | Q6_K | Runs great | 10 GB | 17 | 347 |
| M4 Max · 48 GB | Q6_K | Runs great | 18 GB | 21 | 433 |
| M4 Max · 64 GB | Q6_K | Runs great | 29 GB | 21 | 433 |
| M4 Max · 96 GB | Q6_K | Runs great | 29 GB | 17 | 347 |
| M4 Max · 128 GB | Q6_K | Runs great | 29 GB | 21 | 433 |
| M3 · 8 GB | Q3_K_M | Runs well | 5.0 GB | 6.3 | 72 |
| M3 · 16 GB | Q4_K_M | Runs great | 6.4 GB | 5.5 | 72 |
| M3 · 24 GB | Q6_K | Runs great | 10 GB | 4.6 | 72 |
| M3 Pro · 18 GB | Q6_K | Runs great | 8.2 GB | 6.8 | 126 |
| M3 Pro · 36 GB | Q6_K | Runs great | 10 GB | 6.8 | 126 |
| M3 Max · 36 GB | Q6_K | Runs great | 10 GB | 13 | 269 |
| M3 Max · 48 GB | Q6_K | Runs great | 18 GB | 17 | 359 |
| M3 Max · 64 GB | Q6_K | Runs great | 29 GB | 17 | 359 |
| M3 Max · 96 GB | Q6_K | Runs great | 29 GB | 13 | 269 |
| M3 Max · 128 GB | Q6_K | Runs great | 29 GB | 17 | 359 |
| M3 Ultra · 96 GB | Q6_K | Runs great | 29 GB | 30 | 538 |
| M3 Ultra · 256 GB | Q6_K | Runs great | 29 GB | 30 | 538 |
| M3 Ultra · 512 GB | Q6_K | Runs great | 29 GB | 30 | 718 |
| M2 · 8 GB | Q3_K_M | Runs well | 5.0 GB | 6.3 | 70 |
| M2 · 16 GB | Q4_K_M | Runs great | 6.4 GB | 5.5 | 70 |
| M2 · 24 GB | Q6_K | Runs great | 10 GB | 4.6 | 70 |
| M2 Pro · 16 GB | Q4_K_M | Runs great | 6.4 GB | 11 | 141 |
| M2 Pro · 32 GB | Q6_K | Runs great | 10 GB | 8.9 | 141 |
| M2 Max · 32 GB | Q6_K | Runs great | 10 GB | 17 | 264 |
| M2 Max · 64 GB | Q6_K | Runs great | 29 GB | 17 | 264 |
| M2 Max · 96 GB | Q6_K | Runs great | 29 GB | 17 | 335 |
| M2 Ultra · 64 GB | Q6_K | Runs great | 29 GB | 30 | 528 |
| M2 Ultra · 128 GB | Q6_K | Runs great | 29 GB | 30 | 528 |
| M2 Ultra · 192 GB | Q6_K | Runs great | 29 GB | 30 | 528 |
| M1 · 8 GB | Q3_K_M | Runs well | 5.0 GB | 4.2 | 57 |
| M1 · 16 GB | Q4_K_M | Runs great | 6.4 GB | 3.7 | 57 |
| M1 Pro · 16 GB | Q4_K_M | Runs great | 6.4 GB | 10 | 114 |
| M1 Pro · 32 GB | Q6_K | Runs great | 10 GB | 8.6 | 114 |
| M1 Max · 32 GB | Q6_K | Runs great | 10 GB | 16 | 196 |
| M1 Max · 64 GB | Q6_K | Runs great | 29 GB | 16 | 196 |
| M1 Ultra · 64 GB | Q6_K | Runs great | 29 GB | 26 | 392 |
| M1 Ultra · 128 GB | Q6_K | Runs great | 29 GB | 26 | 392 |
Writes and reads are words a second (a token is about three quarters of a word), 4K tokens into a chat, estimated unless marked measured. The size is the one the store offers that Mac. Constants fitted October 11, 2026; how, and how wrong.
Badge for the model card
Shows on K2-Horizon 7B's Hugging Face card which Macs run it well, linked to this page. It updates every morning.
Sizes and files are read from bartowski/K2-Horizon-7B-GGUF on Hugging Face and checked every morning. LM Arena has not rated this model, so the rating is Outmake's estimate from its size and release date; how the estimate is made. Catalog of November 4, 2026.