[whatmodel.sh](../../README.md) / DeepSeek-Coder-V2 Instruct (0614)
# DeepSeek-Coder-V2 Instruct (0614)
DeepSeek · released 2024-06 · 236B params (21B active) · deepseek
**score 54.0** ±0.6 · #303 of 732 · from 12 benchmark results across 8 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks DeepSeek-Coder-V2 Instruct (0614) among every model on that task or test, worst to best.
task vs the others rank counts tests
coding ▁▁▁▃▅▇●▇▅▄▃▁ #218 of 469 54% 6
chat & writing ▁▁▁▃▆▄●█▆▇▇▃ #267 of 402 19% 3
math ▁▁▃▅●▆▄█▄▂▁▁ #285 of 431* 3.5% 1
coding · #218 of 469 · 54% of its score
Beats [Qwen3.5 4B](qwen3-5-4b.md), [Qwen3-VL 32B Instruct](qwen3-vl-32b-instruct.md), [Solar Open 100B (102B-A12B)](solar-open-100b.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), [s1.1 32B](s1-1-32b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
HumanEval 85.4% ▁·▁▁▃▄▆▅█●▁▁ beats 91/99 easy · avg 76% 18% sharp
HumanEval+ 82.3% ▁·▁▁▁▃▄▄▄█●▁ beats 94/99 easy · avg 70% 17% sharp
LiveCodeBench official (Jan 1–Nov… 34.4% ▁▁·▁▂▄▆●▅▂·▂ beats 19/31 hard · avg 23% 10% sharp · at its level
MBPP+ 75.1% ▁···▁▁▂▅▇█▆● beats 72/75 easy · avg 66% 4.2% at its level
WebApp1K-React pass@1 (onekq-ai) 70.0% ▃▁▁█▁▄▅●▄▇▁▄ beats 18/31 medium · avg 40% 2.1% sharp · at its level · noisy
SciCode main problems (official l… 3.1% ▃···█··●·▃·▂ beats 11/17 very hard · avg 1.0% 1.7% sharp · too hard for it
chat & writing · #267 of 402 · 19% of its score
Beats [Ministral 3 14B Reasoning 2512](ministral-3-14b-reasoning.md), [Ministral 3 14B](ministral-3-14b.md), [OpenThinker2 32B](openthinker2-32b.md) here, though they rank higher overall. Trails [Flan-T5 XXL](flan-t5-xxl.md), [North Micro Vision Instruct (2.4B)](north-micro-vision-instruct.md), [Granite 4.0 H 1B](granite-4-0-h-1b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Arena-Hard (v0.1, GPT-4-1106 judg… 62.3% ▂▁▁▅█▅▄▃●▅▃▁ beats 42/60 medium · avg 47% 10% sharp · at its level
LMArena (text) 1265.4 ▁▁▁▂▃▃●▆▆▇█▃ beats 99/385 avg 1260 4.4% blunt
Polyrating rating (bias-aware re-… 1181.0 ▂▁▂▁▃▇█▆●▄▂▂ beats 76/95 avg 1187 4.1% typical
math · #285 of 431 · 3.5% of its score
Beats [Molmo 2 8B](molmo-2-8b.md), [Llama 3.3 70B Instruct](llama-3-3-70b.md), [DeepSeek-V2.5](deepseek-v2-5.md) here, though they rank higher overall. Trails [DeepSeek R1 Distill Qwen 1.5B](deepseek-r1-distill-qwen-1-5b.md), [Gemma 7B (base)](gemma-7b-base.md), [Mistral 7B v0.1 (base)](mistral-7b-v0-1.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
GSM8K (ZeroEval, zero-shot) 93.8% ▁▄·▁▄█▅▅▅●██ beats 30/43 very easy · avg 87% 3.5% sharp · maxed out
other tests (2)
test result vs the others in its run difficulty counts why
MMLU-Redux (ZeroEval, 2,778 quest… 79.6% ▁·▁▁▄▄▄●█▄▂▁ beats 26/53 easy · avg 75% 16% typical
ZebraLogic (ZeroEval, 1,000 puzzl… 21.1% ▁▁▃▆●█▁▁·▁▁▁ beats 32/57 hard · avg 15% 7.9% sharp · at its level
source results share of the score
EvalPlus / EvalPerf published evaluatio… 3 40% ▇▇▇▇▇
WildEval ZeroEval published results 3 28% ▇▇▇
Arena-Hard-Auto v0.1 leaderboard 1 10% ▇
LiveCodeBench official website results 1 10% ▇
LMArena leaderboard dataset 1 4% ▇
Polyrating: A Cost-Effective and Bias-A… 1 4% ▇
WebApp1K models leaderboard 1 2% ▇
SciCode leaderboard 1 2% ▇
prior: pull toward the middle 0% ▇
weighted average = 54.0
Licences: EvalPlus / EvalPerf published evaluation runs ([Apache-2.0](../../licenses/evalplus-published-runs-Apache-2.0.txt)) · WildEval ZeroEval published results ([Apache-2.0](../../licenses/zeroeval-published-results-Apache-2.0.txt)) · Arena-Hard-Auto v0.1 leaderboard ([Apache-2.0](../../licenses/arena-hard-auto-v0-1-Apache-2.0.txt)) · LiveCodeBench official website results ([CC-BY-SA-4.0](../../licenses/lcb-sharealike-NOTICE.txt)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Polyrating: A Cost-Effective and Bias-Aware Rating System for LLM Evaluation ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · WebApp1K models leaderboard ([MIT](../../licenses/webapp1k-leaderboard-MIT.txt)) · SciCode leaderboard ([Apache-2.0](../../licenses/scicode-Apache-2.0.txt))
## 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-Coder-V2 Instruct (0614) (Q4_K_M, needs ~142.45 GB, from [huggingface.co/bartowski](https://huggingface.co/bartowski/DeepSeek-Coder-V2-Instruct-GGUF/tree/899b543bb21c3a4c146139a7e52c5deb47ad20e3))
```sh
llama cli -hf bartowski/DeepSeek-Coder-V2-Instruct-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-Coder-V2-Instruct-GGUF:Q4_K_M
```
Other versions: Q2_K (~85.95 GB) `llama cli -hf bartowski/DeepSeek-Coder-V2-Instruct-GGUF:Q2_K`
### 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-Coder-V2 Instruct (0614) (Q4_K_M, needs ~142.45 GB, from [huggingface.co/bartowski](https://huggingface.co/bartowski/DeepSeek-Coder-V2-Instruct-GGUF/tree/899b543bb21c3a4c146139a7e52c5deb47ad20e3))
```sh
lms get https://huggingface.co/bartowski/DeepSeek-Coder-V2-Instruct-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: Q2_K (~85.95 GB) `lms get https://huggingface.co/bartowski/DeepSeek-Coder-V2-Instruct-GGUF@Q2_K -y` · 4bit (~132.7 GB, Mac only) `lms get https://huggingface.co/GGorman/DeepSeek-Coder-V2-Instruct-Q4-mlx -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-Coder-V2 Instruct (0614) (4bit, needs ~132.7 GB, Mac only, from [huggingface.co/GGorman](https://huggingface.co/GGorman/DeepSeek-Coder-V2-Instruct-Q4-mlx/tree/26f5b71c868892128eb35b2872664296649af982))
```sh
mlx_lm.chat --model GGorman/DeepSeek-Coder-V2-Instruct-Q4-mlx
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model GGorman/DeepSeek-Coder-V2-Instruct-Q4-mlx
```
Model repository and documentation: [huggingface.co/deepseek-ai](https://huggingface.co/deepseek-ai/DeepSeek-Coder-V2-Instruct)
[interactive version](https://whatmodel.sh/model/deepseek-coder-v2) · [JSON](https://whatmodel.sh/api?format=json&all=1)