text generation Models

Browse open-source large language models in this category.

22 open-source models available

← All Categories
Filters
22 models

LLaMA 3.1 8B

8B

Compact yet capable 8B parameter model. Runs on consumer hardware while maintaining impressive performance.

llamaopen-sourcelocal-friendly
15.7MLlama 3.1 Community License

Mistral 7B v0.3

7B

Efficient 7B parameter model with sliding window attention. Excellent for local deployment.

mistralopen-sourcelocal-friendly
9.8MApache 2.0

LLaMA 3.1 70B

70B

Meta's 70B parameter model offering excellent performance with lower resource requirements than 405B.

llamaopen-sourcereasoning
8.9MLlama 3.1 Community License

Qwen 2.5 7B

7B

Compact Qwen model with strong performance for its size. Great for local deployment.

qwenopen-sourcelocal-friendly
7.7MApache 2.0

Gemma 2 2B

2B

Ultra-lightweight 2B model. Runs on phones and edge devices.

gemmaopen-sourcegoogle
6.8MGemma Terms of Use

Phi 3.5 Mini

3.8B

Microsoft's 3.8B parameter model that punches far above its weight class.

phiopen-sourcetiny-but-mighty
5.7MMIT

Gemma 2 9B

9B

Efficient 9B model from Google with strong reasoning. Fits on most modern GPUs.

gemmaopen-sourcegoogle
4.6MGemma Terms of Use

LLaMA 3.2 3B

3B

Meta's lightweight 3B model for edge and mobile deployment.

llamaopen-sourceedge
4.6MLlama 3.2 Community License

Qwen 2.5 72B

72B

Alibaba's flagship 72B model with exceptional coding and mathematical reasoning.

qwenopen-sourcecode
3.5MApache 2.0

LLaMA 3.1 405B

405B

Meta's largest open model with 405B parameters. State-of-the-art performance across reasoning, code, and multilingual tasks.

llamaopen-sourcereasoning
2.8MLlama 3.1 Community License

Gemma 2 27B

27B

Google's powerful 27B model with knowledge distillation from Gemini. Excellent quality.

gemmaopen-sourcegoogle
2.3MGemma Terms of Use

Phi 3 Medium

14B

Microsoft's 14B parameter model with excellent reasoning capabilities.

phiopen-sourcereasoning
2.3MMIT

Mixtral 8x22B

141B

Sparse mixture-of-experts model with 141B total / 39B active parameters. Outstanding efficiency.

mistralopen-sourcemoe
1.9MApache 2.0

DeepSeek V3

671B

DeepSeek's 671B MoE model with 37B active parameters. Matches GPT-4o on many benchmarks.

deepseekopen-sourcemoe
1.6MDeepSeek License

Mistral Large 2

123B

Mistral's flagship 123B parameter model with strong multilingual and coding capabilities.

mistralopen-sourcemultilingual
1.2MMistral Research License

SmolLM2 1.7B

1.7B

HuggingFace's ultra-compact 1.7B model. Best-in-class for its size.

smollmopen-sourcetiny
1.2MApache 2.0

Yi 1.5 34B

34B

01.AI's 34B parameter model with strong bilingual (EN/CN) capabilities.

yiopen-sourcebilingual
877KApache 2.0

WizardMath 70B

70B

Math-specialized LLaMA 2 70B fine-tuned with RLHF for mathematical reasoning.

llamaopen-sourcemath
765KLlama 2 Community License

Command R+

104B

Cohere's 104B parameter model optimized for RAG and tool use.

command-ropen-sourcerag
654KCC-BY-NC-4.0

Command R

35B

Cohere's 35B model for RAG, summarization, and tool use.

command-ropen-sourcerag
543KCC-BY-NC-4.0

Falcon 180B

180B

TII's massive 180B model trained on 3.5T tokens of RefinedWeb data.

falconopen-sourcemassive
432KFalcon-180B TII License

Granite 3.1 8B

8B

IBM's enterprise-grade 8B model with strong reasoning and code capabilities.

graniteopen-sourceenterprise
346KApache 2.0