T-Pro or GigaChat: which one for a Russian company
Both lines are Russian, both list Russian language support and both are open for commercial use: T-Pro under Apache 2.0, GigaChat under MIT. T-Pro and T-Lite from T-Bank are fine-tuned on top of Qwen, stay in the 7B to 36B-A3B range and expect a GPU, while T-Search 36B-A3B acts as a multi-step search agent. GigaChat from Sber covers a wider range, 10B-A1.8B to 702B-A36B, runs on a CPU as well, and GigaChat3.1-Audio handles recordings up to two hours. Neither is in Ollama, so installation is manual; GigaChat was updated in September 2026, T-Pro in July 2026.
Comparison based on catalog data
| Parameter | T-Pro / T-Lite | GigaChat |
|---|---|---|
| Category | Text | Text, Voice assistants |
| Developer | T-Bank, Russia | Sber, Russia |
| Releases | Dec 2024 – Jul 2026 | Dec 2024 – Sep 2026 |
| Sizes | 7B – 36B-A3B | 10B-A1.8B – 702B-A36B |
| Hardware | Laptop, 1 GPU | Laptop, 1 GPU, Cluster |
| Commercial use | Commercial use allowed | Commercial use allowed |
| License | Apache 2.0 (versions 2.0, 2.1 and T-Search; based on Qwen) | MIT |
| Russian | Supported | Supported |
| Ollama | No | No |
| Without GPU | No | Yes |
| Tasks |
|
|
Choose T-Pro / T-Lite if
- You need a single-GPU model: T-Pro is 32B
- Your work is banking, fintech, legal or HR in Russian
- You need a multi-step search agent: T-Search 36B-A3B
Choose GigaChat if
- You are a public body or must keep data inside Russia
- You need audio: GigaChat3.1-Audio takes recordings up to two hours
- Hardware varies: from 10B-A1.8B on a CPU to 702B on a cluster
Other comparisons
- Qwen or GigaChat: which to choose for business
- GigaChat or YandexGPT: which to choose for business
- Qwen or Llama: which to choose for business
- DeepSeek or Qwen: which to choose for business
- Gemma or Llama: which to choose for business
- Mistral or Llama: which to choose for business
- gpt-oss or Qwen: which one to run on your own server
- Llama or DeepSeek: which to run inside your own perimeter
- Mistral or Qwen: which to choose for business
- Gemma or Phi: small models for modest hardware
- YandexGPT or Qwen: which to choose for Russian
- DeepSeek-R1 or QwQ: which reasoning model to pick
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