Best open-weight LLMs
Current models whose weights you can download, sorted by capability tier, with licenses and sizes spelled out.
Open weights give you control: run the model in your own cloud or data center, fine-tune it, and keep sensitive data inside your network. You can also buy the same model from several inference providers, which keeps prices competitive.
Open weights don't always mean open source. Some licenses cap revenue, require attribution, restrict serving the model to others or forbid certain uses. Read the license on the model page before you build a product on it.
The shortlist
| Model | Tier | Weights | Context | Input / 1M | Output / 1M | Inputs |
|---|---|---|---|---|---|---|
| DeepSeek-V4-ProDeepSeek | Frontier | Open weights | 1M | $1.32 | $3.96 | Text |
| GLM-5.3Z.ai (Zhipu AI) | Frontier | Open weights | 1M | $1.40 | $4.40 | Text |
| Qwen3.8-2.4T-A95BAlibaba (Qwen) | Frontier | Open weights | 256K | $2 | $6 | Text |
| Kimi K3Moonshot AI | Frontier | Open weights | 1M | $3 | $15 | Text, Images, Video |
| gpt-oss-120bOpenAI | Advanced | Open weights | 128K | $0.15 | $0.60 | Text |
| DeepSeek-V4.1-FlashDeepSeek | Advanced | Open weights | 1M | $0.30 | $1.20 | Text, Images |
| MiniMax-M3MiniMax | Advanced | Open weights | 1M | $0.30 | $1.20 | Text, Images, Video |
| Mistral Large 3Mistral AI | Advanced | Open weights | 256K | $0.50 | $1.50 | Text, Images |
| NVIDIA Nemotron 3 UltraNVIDIA | Advanced | Open weights | 1M | $0.50 | $2.20 | Text |
| Mistral Medium 3.5Mistral AI | Advanced | Open weights | 256K | $1.50 | $7.50 | Text, Images |
| Ministral 3 14BMistral AI | Efficient | Open weights | 256K | $0.20 | $0.20 | Text, Images |
| GLM-5.3-FlashZ.ai (Zhipu AI) | Efficient | Open weights | 1M | $0.15 | $0.50 | Text, Images, Video |
| Gemma 4 31BGoogle | Efficient | Open weights | 256K | $0.20 | $0.40 | Text, Images, Video |
| Mistral Small 4Mistral AI | Efficient | Open weights | 256K | $0.15 | $0.60 | Text, Images |
| Granite 4.2 30BIBM | Efficient | Open weights | 128K | $0.16 | $0.65 | Text |
| Muse Glimmer 30BMeta | Efficient | Open weights | 128K | $0.35 | $1.50 | Text, Images, Video |
| Qwen3.8-27BAlibaba (Qwen) | Efficient | Open weights | 256K | $0.80 | $4 | Text, Images, Video |
| Olmo 3.1 32B ThinkAi2 (Allen Institute for AI) | Efficient | Open weights | 64K | — | — | Text |
| NVIDIA Nemotron 3.5 Lightning 30B-A3BNVIDIA | Compact | Open weights | 1M | $0.05 | $0.20 | Text |
| Ministral 3 3BMistral AI | Compact | Open weights | 256K | $0.10 | $0.10 | Text, Images |
| Granite 4.2 8BIBM | Compact | Open weights | 128K | $0.06 | $0.25 | Text |
| gpt-oss-20bOpenAI | Compact | Open weights | 128K | $0.075 | $0.30 | Text |
| Gemma 4 26B A4BGoogle | Compact | Open weights | 256K | $0.07 | $0.34 | Text, Images, Video |
| Ministral 3 8BMistral AI | Compact | Open weights | 256K | $0.15 | $0.15 | Text, Images |
| Qwen3.6-35B-A3BAlibaba (Qwen) | Compact | Open weights | 256K | $0.10 | $0.95 | Text, Images, Video |
| Gemma 4 E4BGoogle | Compact | Open weights | 128K | — | — | Text, Images, Audio, Video |
| Phi-4-reasoning-vision-15BMicrosoft (Phi) | Compact | Open weights | 16K | — | — | Text, Images |
| Phi-4-mini-instructMicrosoft (Phi) | Compact | Open weights | 128K | — | — | Text |
Ordered by status, capability tier and then price. Tiers are our editorial grouping of provider positioning and public benchmark standings, not our own test results. How we rank.
How to choose
- Match model size to your hardware: the weights alone need about 0.6 GB per billion parameters at 4-bit.
- Check the license for commercial limits, especially for the largest releases.
- Compare hosted prices for the same open model across inference providers.
- Read open-source vs closed LLMs for the full trade-off.
Frequently asked questions
What is the best open-weight LLM?
The frontier-tier models at the top of this list are the most capable open releases we track, but they need multi-GPU servers. For a single GPU or a laptop, the efficient and compact models are the practical choice. The best one is the most capable model that fits your hardware and license needs.
Are open-weight models free to use commercially?
Many are, under licenses such as Apache 2.0 or MIT. Others use custom licenses with revenue caps, attribution rules or extra terms for large service providers. The license name is listed for every model here; read the full text before commercial use.
Other shortlists
Planning to self-host?
Get a second opinion on hardware, serving stack and break-even before you buy or rent GPUs.
- Free and vendor-neutral: no model provider pays for placement.
- Your details go only to partners you agree to.