LLM
GA
Open weights
Qwen3.8-2.4T-A95B
The largest open-weight frontier-class model
Release2026-09
Context1M
Max output—
Parameters2.4T
Local / VRAM2600–4000 GB
LicenseOpen weights
Overview
Qwen3.8-2.4T-A95B is a model in the LLM category from Alibaba (Qwen). It was released in 2026-09. The vendor is based in China. Best for: The largest open-weight frontier-class model.
Hardware: Multi-B200/H200 cluster; cloud-only in practice.
Strengths
- 2.4T MoE
- 1M context
- Open weights
Limits
- Self-hosting is impractical
- Conditional license
Specifications
| Vendor | Alibaba (Qwen) |
|---|---|
| Category | LLM |
| Release | 2026-09 |
| Status | GA |
| License | Open weights |
| License name | Qwen3.8-Max License |
| Commercial use | Conditional |
| Parameters | 2.4T |
| Active parameters | 95B |
| Architecture | MoE transformer |
| Context window | 1M |
| Input modalities | Text, Image |
| Output modalities | Text |
| Prompt types | Text |
| Negative prompts | No |
| Max outputs | 1 |
| Local deployment | Yes |
| VRAM | 2600–4000 GB |
| System RAM | 3000–6000 GB |
| API access | Yes |
| Pricing | Free weights |
| Confidence | Medium |
| Vendor country | China |
| Vendor type | Cloud provider |
Pricing & hardware
Free weights
This model can be self-hosted or run locally. Approximate GPU memory: 2600–4000 GB. Approximate system RAM: 3000–6000 GB. API access is available.