Available Local LLMs

Find reference information and an up-to-date (January 31st, 2025) of local LLMs available for download that are currently supported by PiecesOS, the Pieces Desktop App, and other Pieces plugins and extensions.


Supported LLMs

The Pieces for Developers Suite currently supports 41 local models from a range of providers.


Provider

Model Name

Google

Gemma / Code Gemma

IBM

Granite / Code / Dense / MoE

Meta

LLaMA / CodeLLaMA

Mistral

Mistral / Mixtral

Microsoft

Phi

Qwen

QwQ / Coder

StarCoder

StarCoder

View the tables below for detailed model names, parameters, and the context window size of all usable models.


Please note that not all specific models have easily indentifiable parameter quantities. Some companies release information on their models, while others do not—as such, the parameters provided in these tables are estimated parameter ranges based on leading AI sources, detailed evaluations and assessments, and other available information.


Google (Gemma)


Model Name

Parameters

Context Window (Maximum)

Gemma 2 27B

27B

8k tokens

Gemma 2 9B

9B

8k tokens

Gemma 2 2B

2B

8k tokens

Gemma 1.1 7B

7B

4k tokens

Gemma 1.1 2B

2B

4k tokens

Code Gemma 1.1 7B

7B

4k tokens

IBM (Granite)


Model Name

Parameters

Context Window (Maximum)

Granite Code 34B

34B

8k tokens

Granite Code 20B

20B

8k tokens

Granite Code 8B

8B

128k tokens

Granite Code 3B 128K

3B

128k tokens

Granite Code 3B

3B

4k tokens

Granite 3.1 Dense 8B

8B

128k tokens

Granite 3.1 Dense 2B

2B

128k tokens

Granite 3 MoE 3B

3B

128k tokens

Granite 3 MoE 1B

1B

128k tokens

Granite 3 Dense 8B

8B

128k tokens

Meta (LLaMA)


Model Name

Parameters

Context Window (Maximum)

LLaMA 3.2 3B

3B

128k tokens

LLaMA 3.2 1B

1B

8k tokens

LLaMA 3 8B

8B

8k tokens

LLaMA 2 13B

13B

4lk tokens

LLaMA 2 7B

7B

4k tokens

CodeLLaMA 34B

34B

100k tokens

CodeLLaMA 13B

13B

16k tokens

CodeLLaMA 7B

7B

8k tokens

Mistral (Mixtral)


Model Name

Parameters

Context Window (Maximum)

Mixtral 8 7B

7B

128k tokens

Mistral 7B

7B

32.8k tokens

Microsoft (Phi)


Model Name

Parameters

Context Window

Phi-4 14B

14B

4k tokens

Phi-3.5 Mini 3.8B

3.8B

128k tokens

Phi-3 Mini 128K

3B

128k tokens

Phi-3 Mini 4K

3B

4k tokens

Phi-3 Medium 14B 128K

14B

128k tokens

Phi-3 Medium 14B 4K

14B

4k tokens

Phi-2

N/A

4k tokens

Qwen (Qwen)


Model Name

Parameters

Context Window

Qwen QwQ Preview 32B

32B

32k tokens

Qwen 2.5 Coder 32B

32B

128k tokens

Qwen 2.5 Coder 14B

14B

32k tokens

Qwen 2.5 Coder 7B

7B

128k tokens

Qwen 2.5 Coder 3B

3B

32k tokens

Qwen 2.5 Coder 1.5B

1.5B

128k tokens

Qwen 2.5 Coder 0.5B

0.5B

32k tokens

StarCoder (StarCoder)


Model Name

Parameters

Context Window

StarCode 2 15B

15b

16k tokens

Updated on