gpt-oss or Qwen: which one to run on your own server
Both families install through Ollama and run in a closed environment with no cloud. gpt-oss comes in two sizes, 20B and 120B, entirely under Apache 2.0 with unconditional commercial use, but it needs a GPU. Qwen starts at 0.6B, runs on a CPU and is listed in the catalog as supporting Russian, though the larger Qwen3.8 models carry their own license and conditional commercial terms. On dates, the catalog has a single gpt-oss release from August 2025, while Qwen versions continue through August 2026.
Comparison based on catalog data
| Parameter | gpt-oss | Qwen |
|---|---|---|
| Category | Text | Text, Image + text |
| Developer | OpenAI, USA | Alibaba, China |
| Releases | Aug 2025 – Aug 2025 | Aug 2023 – Aug 2026 |
| Sizes | 20B, 120B | 0,6B – 2,4T-A95B |
| Hardware | 1 GPU | Laptop, 1 GPU, Cluster |
| Commercial use | Commercial use allowed | Commercial use with conditions |
| License | Apache 2.0 | Apache 2.0 (most versions); the larger Qwen3.8 models have their own license |
| Russian | Not stated | Supported |
| Ollama | Yes | Yes |
| Without GPU | No | Yes |
| Tasks |
|
|
Choose gpt-oss if
- You want plain Apache 2.0 with no strings attached to commercial use
- You are building a corporate agent that calls internal systems
- You have a GPU: the smaller gpt-oss-20b fits on one card
Choose Qwen if
- Your assistant must answer in Russian, which Qwen lists and gpt-oss does not
- No GPU available: Qwen starts at 0.6B and runs on a CPU
- You want current releases and sizes all the way up to 2.4T-A95B
Other comparisons
- Qwen or GigaChat: which to choose for business
- Qwen or Llama: which to choose for business
- DeepSeek or Qwen: which to choose for business
- Mistral or Qwen: which to choose for business
- YandexGPT or Qwen: which to choose for Russian
- GLM or Qwen: which to choose for agents and documents
- GigaChat or YandexGPT: which to choose for business
- Gemma or Llama: which to choose for business
- Mistral or Llama: which to choose for business
- Llama or DeepSeek: which to run inside your own perimeter
- Gemma or Phi: small models for modest hardware
- T-Pro or GigaChat: which one for a Russian company
Need a model for your task?
An open model can run on your own server: data stays in-house, there is no per-request fee, and the model can be fine-tuned on your documents.
- SelectThe model and size for your task and hardware budget
- DeployOn your server or in a closed network, with an API
- Fine-tuneOn your data, or connect a knowledge base
- IntegrateInto your CRM, ERP, bot, website or team chat