[whatmodel.sh](../../README.md) / DeepSeek-V2.5 (1210)
# DeepSeek-V2.5 (1210)
DeepSeek · released 2024-12 · 236B params (21B active) · deepseek
**score 56.4** ±1.3 · #253 of 732 · from 6 benchmark results across 5 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks DeepSeek-V2.5 (1210) among every model on that task or test, worst to best.
task vs the others rank counts tests
coding ▁▁▁▃▅▇█●▅▄▃▁ #208 of 469* 66% 2
chat & writing ▁▁▁▃▆▄▆●▆▇▇▃ #199 of 402* 27% 2
math ▁▁▃▅█●▄█▄▂▁▁ #240 of 431* 5.5% 1
coding · #208 of 469 · 66% of its score
Beats [Qwen3 30B-A3B Instruct 2507](qwen3-30b-a3b-instruct-2507.md), [Qwen3.5 4B](qwen3-5-4b.md), [Falcon-H1R 7B](falcon-h1r-7b.md) here, though they rank higher overall. Trails [QwQ-32B-Preview](qwq-32b-preview.md), [Ministral 3 3B Reasoning 2512](ministral-3-3b-reasoning.md), [DeepSeek-Coder-V2 Instruct (0724)](deepseek-coder-v2-0724.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Aider polyglot (225 exercises, tw… 17.8% ▂·▂●▅▂▃█▆▄▂▂ beats 8/49 very hard · avg 7.6% 35% sharp
Aider code editing (133 Exercism … 72.2% ▁·▁▂▄▁▆▄█●▂▂ beats 53/62 medium · avg 51% 31% sharp · at its level
chat & writing · #199 of 402 · 27% of its score
Beats [Gemma 2 9B IT SimPO](gemma-2-9b-simpo.md), [Granite 4.2 8B](granite-4-2-8b.md), [Mistral Small 4 (119B-A6.5B)](mistral-small-4.md) here, though they rank higher overall. Trails [LFM2.5 1.2B Instruct](lfm2-5-1-2b-instruct.md), [LFM2.5 1.2B Thinking](lfm2-5-1-2b-thinking.md), [LFM2.5 8B-A1B](lfm2-5-8b-a1b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LMArena (text) 1323.4 ▁▁▁▂▃▃▃●▆▇█▃ beats 159/385 avg 1260 18% blunt
IFEval prompt-level strict 79.3% ▄●··▄·▄··█·▄ beats 1/6 easy · avg 73% 8.4% typical
math · #240 of 431 · 5.5% of its score
Beats [GPT-4o (2024-08-06)](gpt-4o-2024-08-06.md), [Command A (03-2025)](command-a.md), [GPT-4o (2024-05-13)](gpt-4o-2024-05-13.md) here, though they rank higher overall. Trails [Gemma 7B (base)](gemma-7b-base.md), [CodeGemma 7B (v1.0)](codegemma-7b.md), [Phi-2 (2.7B)](phi-2.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
MATH-500 (maker) 82.8% ▁▃▂▃▄●▅▇█▄▂▁ beats 24/85 easy · avg 76% 5.5% small run
other tests (1)
test result vs the others in its run difficulty counts why
DROP F1, 3-shot (simple-evals) 89.1% ▅▂▆█▆·▃·●▅·▂ beats 19/23 easy · avg 81% 1.0% blunt · small run
source results share of the score
Aider polyglot benchmark runs 1 35% ▇▇▇▇
Aider code editing benchmark runs 1 31% ▇▇▇▇
LMArena leaderboard dataset 1 18% ▇▇
OpenCompass CompassAcademic leaderboard 2 9% ▇
DeepSeek vendor-reported benchmark scor… 1 6% ▇
prior: pull toward the middle 1% ▇
weighted average = 56.4
Licences: Aider polyglot benchmark runs ([Apache-2.0](../../licenses/aider-polyglot-runs-Apache-2.0.txt)) · Aider code editing benchmark runs ([Apache-2.0](../../licenses/aider-polyglot-runs-Apache-2.0.txt)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · OpenCompass CompassAcademic leaderboard ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0))
## How to run it
### llama.cpp (recommended)
1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md)
```sh
# macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
# Windows
irm https://llama.app/install.ps1 | iex
```
2. Run DeepSeek-V2.5 (1210) (Q4_K_M, needs ~142.45 GB, from [huggingface.co/bartowski](https://huggingface.co/bartowski/DeepSeek-V2.5-1210-GGUF/tree/fb3b91fa0262ceae59a8055b297497cddabdbcdf))
```sh
llama cli -hf bartowski/DeepSeek-V2.5-1210-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf bartowski/DeepSeek-V2.5-1210-GGUF:Q4_K_M
```
Other versions: IQ2_M (~76.92 GB) `llama cli -hf bartowski/DeepSeek-V2.5-1210-GGUF:IQ2_M`
### LM Studio
1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download)
2. Run DeepSeek-V2.5 (1210) (Q4_K_M, needs ~142.45 GB, from [huggingface.co/bartowski](https://huggingface.co/bartowski/DeepSeek-V2.5-1210-GGUF/tree/fb3b91fa0262ceae59a8055b297497cddabdbcdf))
```sh
lms get https://huggingface.co/bartowski/DeepSeek-V2.5-1210-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: IQ2_M (~76.92 GB) `lms get https://huggingface.co/bartowski/DeepSeek-V2.5-1210-GGUF@IQ2_M -y` · 4bit (~132.7 GB, Mac only) `lms get https://huggingface.co/mlx-community/DeepSeek-V2.5-1210-4bit -y` · 8bit (~250.5 GB, Mac only) `lms get https://huggingface.co/dong-99/DeepSeek-V2.5-1210-mlx-8Bit -y`
### MLX (Mac)
1. Install MLX (Mac). Apple silicon Macs only, text only.
```sh
# with uv
uv tool install mlx-lm
# or
pip install mlx-lm
```
2. Run DeepSeek-V2.5 (1210) (4bit, needs ~132.7 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/DeepSeek-V2.5-1210-4bit/tree/3b6ce93c9c4cb6d390377d183f66aaf5e3b4f507))
```sh
mlx_lm.chat --model mlx-community/DeepSeek-V2.5-1210-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model mlx-community/DeepSeek-V2.5-1210-4bit
```
Other versions: 8bit (~250.5 GB, Mac only) `mlx_lm.chat --model dong-99/DeepSeek-V2.5-1210-mlx-8Bit`
Model repository and documentation: [huggingface.co/deepseek-ai](https://huggingface.co/deepseek-ai/DeepSeek-V2.5-1210)
[interactive version](https://whatmodel.sh/model/deepseek-v2-5-1210) · [JSON](https://whatmodel.sh/api?format=json&all=1)