Comparisons of Small Open Source AI Models (4B-40B)
Open source AI models with less than 40 billion parameters. Models are considered Open Source (also commonly referred to as open weights) where their weights are accessible to download. This allows self-hosting on your own infrastructure and enables customizing the model such as through fine-tuning. Click on any model to see detailed metrics. For more details including relating to our methodology, see our FAQs.
Qwen3 30B 2507 (Reasoning) and
EXAONE 4.0 32B (Reasoning) are the highest intelligence Small open source models, defined as those with 4B-40B parameters, followed by
gpt-oss-20B (high) &
Qwen3 30B 2507 (Non-reasoning).
Highlights
Intelligence
Artificial Analysis Intelligence Index; Higher is better
Estimate (independent evaluation forthcoming)
Total Parameters
Trainable parameters in billions
Further details
Weights | Provider Benchmarks | |||||||
---|---|---|---|---|---|---|---|---|
Qwen3 30B A3B 2507 (Reasoning) Alibaba | 54 | 30.5B (3.3B active at inference time) | 262k | $0.8 | - | 🤗 | ![]() | View |
![]() EXAONE 4.0 32B (Reasoning) LG AI Research | 51 | 32B | 131k | $0.7 | 80 | 🤗 | View | |
gpt-oss-20B OpenAI | 49 | 21B (3.6B active at inference time) | 131k | $0.1 | 260 | 🤗 | ![]() +8 more | View |
Qwen3 30B A3B 2507 (Non-reasoning) Alibaba | 44 | 30.5B (3.3B active at inference time) | 262k | $0.3 | - | 🤗 | ![]() | View |
Qwen3 32B (Reasoning) Alibaba | 44 | 32.8B | 32.8k | $2.6 | - | 🤗 | ![]() ![]() +6 more | View |
NVIDIA Nemotron Nano 9B V2 (Reasoning) NVIDIA | 43 | 9B | 131k | - | - | Not available | - | View |
Qwen3 30B A3B (Reasoning) Alibaba | 42 | 30.5B (3.3B active at inference time) | 32.8k | $0.8 | - | 🤗 | ![]() +4 more | View |
NVIDIA Nemotron Nano 9B V2 NVIDIA | 42 | 9B | 131k | - | - | Not available | - | View |
QwQ 32B Alibaba | 42 | 32.8B | 131k | $0.7 | 57 | 🤗 | +4 more | View |
Qwen3 8B (Reasoning) Alibaba | 41 | 8.19B | 131k | $0.7 | - | 🤗 | View | |
Qwen3 14B (Reasoning) Alibaba | 40 | 14.8B | 32.8k | $1.3 | - | 🤗 | ![]() ![]() | View |
DeepSeek R1 0528 Qwen3 8B DeepSeek | 39 | 8.19B | 32.8k | $0.1 | 81 | 🤗 | View | |
![]() EXAONE 4.0 32B LG AI Research | 37 | 32B | 131k | $0.7 | 75 | Not available | View | |
DeepSeek R1 Distill Qwen 32B DeepSeek | 36 | 32B | 128k | $0.2 | 20 | 🤗 | ![]() | View |
Reka Flash 3 Reka AI | 36 | 21B | 128k | $0.3 | - | 🤗 | View | |
![]() Magistral Small Mistral | 36 | 23.6B | 40.0k | $0.8 | 187 | 🤗 | ![]() | View |
Qwen3 Coder 30B A3B Alibaba | 36 | 30.5B (3.3B active at inference time) | 262k | $0.9 | - | 🤗 | ![]() | View |
DeepSeek R1 Distill Qwen 14B DeepSeek | 33 | 14B | 128k | $0.9 | 43 | 🤗 | View | |
Qwen3 14B Alibaba | 32 | 14.8B | 32.8k | $0.6 | - | 🤗 | ![]() | View |
![]() Mistral Small 3.2 Mistral | 32 | 24B | 128k | $0.1 | 150 | 🤗 | ![]() ![]() | View |
QwQ 32B-Preview Alibaba | 32 | 32.8B | 32.8k | $0.7 | 54 | 🤗 | ![]() | View |
Qwen3 30B A3B Alibaba | 30 | 30.5B (3.3B active at inference time) | 32.8k | $0.3 | - | 🤗 | ![]() | View |
Qwen3 32B Alibaba | 30 | 32.8B | 32.8k | $1.2 | - | 🤗 | ![]() ![]() +7 more | View |
Llama 3.1 Nemotron Nano 4B v1.1 (Reasoning) NVIDIA | 30 | 4.51B | 128k | - | - | 🤗 | - | View |
Qwen3 4B (Reasoning) Alibaba | 29 | 4.02B | 32.0k | $0.4 | - | 🤗 | ![]() | View |
Phi-4 Microsoft Azure | 28 | 14B | 16.0k | $0.2 | 36 | 🤗 | ![]() ![]() | View |
![]() Mistral Small 3.1 Mistral | 26 | 24B | 128k | $0.1 | 164 | 🤗 | ![]() +3 more | View |
Qwen2.5 Instruct 32B Alibaba | 26 | 32B | 128k | $0.1 | - | 🤗 | ![]() ![]() | View |
Qwen3 8B Alibaba | 25 | 8.19B | 32.8k | $0.3 | - | 🤗 | View | |
Gemma 3 27B Instruct Google | 25 | 27.4B | 128k | - | 48 | 🤗 | ![]() | View |
Qwen2.5 Coder Instruct 32B Alibaba | 25 | 32B | 131k | $0.1 | 46 | 🤗 | +1 more | View |
![]() Mistral Small 3 Mistral | 24 | 24B | 32.0k | $0.1 | 135 | 🤗 | ![]() ![]() | View |
Gemma 3 12B Instruct Google | 24 | 12.2B | 128k | $0.2 | - | 🤗 | ![]() | View |
Qwen3 4B Alibaba | 24 | 4.02B | 32.0k | $0.2 | - | 🤗 | View | |
DeepSeek R1 Distill Llama 8B DeepSeek | 23 | 8B | 128k | $0.0 | 50 | 🤗 | View | |
![]() Devstral Small (May '25) Mistral | 23 | 23.6B | 256k | $0.1 | 104 | 🤗 | ![]() ![]() | View |
Solar Mini Upstage | 22 | 10.7B | 4.10k | $0.1 | 81 | 🤗 | View | |
![]() Devstral Small (Jul '25) Mistral | 21 | 24B | 256k | $0.1 | 150 | 🤗 | ![]() ![]() ![]() | View |
Gemma 2 27B Google | 20 | 27.2B | 8.19k | $0.8 | - | 🤗 | View | |
Llama 3.1 Instruct 8B Meta | 19 | 8B | 128k | $0.1 | 167 | 🤗 | ? +17 more | View |
Gemma 3n E4B Instruct Google | 18 | 8.39B (4.0B active at inference time) | 32.0k | $0.0 | 71 | 🤗 | View | |
![]() DeepHermes 3 - Mistral 24B Preview Nous Research | 18 | 24B | 32.0k | - | - | 🤗 | - | View |
Gemma 3 4B Instruct Google | 18 | 4.3B | 128k | $0.0 | - | 🤗 | ![]() | View |
Granite 3.3 Instruct 8B IBM | 18 | 8.17B | 128k | $0.1 | 104 | 🤗 | ![]() | View |
Llama 3.2 Instruct 11B (Vision) Meta | 17 | 11B | 128k | $0.2 | 72 | 🤗 | ![]() | View |
![]() Mistral Small (Sep '24) Mistral | 16 | 22B | 32.8k | $0.3 | 91 | 🤗 | ![]() | View |
Gemma 3n E4B Instruct Preview (May '25) Google | 15 | 8.39B (4B active at inference time) | 32.0k | - | - | 🤗 | - | View |
Phi-4 Multimodal Instruct Microsoft Azure | 15 | 5.6B | 128k | - | 18 | 🤗 | View | |
Qwen2.5 Coder Instruct 7B Alibaba | 15 | 7.62B | 131k | - | - | 🤗 | - | View |
Llama 2 Chat 7B Meta | 14 | 7B | 4.10k | $0.1 | 121 | 🤗 | ![]() | View |
Phi-3 Medium Instruct 14B Microsoft Azure | 13 | 14B | 128k | $0.3 | 43 | 🤗 | View | |
![]() Pixtral 12B (2409) Mistral | 11 | 12B | 128k | $0.1 | 106 | 🤗 | ![]() | View |
Gemma 3n E2B Instruct Google | 10 | 5.98B (2.0B active at inference time) | 32.0k | - | 43 | 🤗 | View | |
![]() Ministral 8B Mistral | 10 | 8B | 128k | $0.1 | 177 | 🤗 | ![]() | View |
Gemma 2 9B Google | 10 | 9B | 8.19k | $0.2 | - | 🤗 | ![]() ![]() | View |
Llama 3 Instruct 8B Meta | 9 | 8B | 8.19k | $0.1 | 73 | 🤗 | ![]() ? +4 more | View |
DeepSeek Coder V2 Lite Instruct DeepSeek | 8 | 16B (2.4B active at inference time) | 128k | - | - | 🤗 | - | View |
![]() Codestral (May '24) Mistral | 8 | 22B | 32.8k | $0.3 | - | 🤗 | ![]() | View |
![]() Aya Expanse 32B Cohere | 8 | 32B | 128k | $0.8 | 100 | 🤗 | ![]() | View |
DeepSeek LLM 67B Chat (V1) DeepSeek | 8 | 7B | 4.10k | - | - | 🤗 | - | View |
Llama 2 Chat 13B Meta | 8 | 13B | 4.10k | - | - | 🤗 | - | View |
![]() OpenChat 3.5 (1210) OpenChat | 8 | 7B | 8.19k | - | - | 🤗 | - | View |
![]() Mistral NeMo Mistral | 8 | 12B | 128k | $0.1 | 113 | 🤗 | ![]() ![]() +2 more | View |
![]() DeepHermes 3 - Llama-3.1 8B Preview Nous Research | 4 | 8B | 128k | - | - | 🤗 | - | View |
![]() Aya Expanse 8B Cohere | 4 | 8B | 8.00k | $0.8 | 136 | 🤗 | ![]() | View |
![]() Command-R (Aug '24) Cohere | 3 | 32B | 128k | $0.3 | 84 | 🤗 | ![]() | View |
![]() Command-R (Mar '24) Cohere | 2 | 35B | 128k | $0.8 | 85 | 🤗 | ![]() | View |
![]() Codestral-Mamba Mistral | 2 | 7B | 256k | - | - | 🤗 | - | View |
![]() Mistral 7B Instruct Mistral | 1 | 7B | 8.19k | $0.3 | 102 | 🤗 | ? +3 more | View |