// Qwen · Text LLM
Qwen3.8-2.4T-A95B: VRAM & local deployment
2.4T total · 95B active (MoE) parameters. About 1.60 TB at 4-bit and 5.28 TB at BF16 with one 8K-token request.
released 2026-08-08licence: Custom licence – read itmemory figures are estimates
01Specs
| Publisher | Alibaba Qwen · Qwen |
|---|---|
| Family / series | Qwen · Qwen3.8 |
| Type | Text LLM (text-generation) |
| Parameters | 2.4T total · 95B active (MoE) Source: official model card“Number of Parameters: 2.4T in total and 95B activated” |
| Context length | 256K tokens (262,144)max_position_embeddings in config.json |
| Architecture | 92 layers (23 full-attention) · GQA+HYBRID · MoE |
| Released weights | BF16 · 4.89 TBdownload |
| Licence | qwen3.8-max · Custom licence – read itThe publisher uses its own licence; we have not classified it. Read the licence text. |
| Released | 2026-08-08 (Hugging Face repo created) |
02How much VRAM it needs (estimate)
Total = weights + 10% + 0.5GB overhead + KV cache for one request (FP16 KV). More concurrent users multiply the KV column.
| Precision | Weights | Total @ 8K | Total @ 32K | Total @ 128K | Smallest GPU | Mac |
|---|---|---|---|---|---|---|
| BF16 / FP16 (original) | 4.80 TB | 5.28 TB | 5.28 TB | 5.29 TB | multi-node | — |
| FP8 / INT8 / Q8_0 | 2.55 TB | 2.81 TB | 2.81 TB | 2.82 TB | 2 × 8 × B300 nodes (4,608GB) | — |
| 4-bit (Q4_K_M / AWQ / GPTQ-Int4) | 1.46 TB | 1.60 TB | 1.60 TB | 1.61 TB | 8 × B300 (2,304GB) | — |
KV cache per request: 8K 0.8 GB · 32K 3.1 GB · 128K 12.3 GB
Bits per weight: BF16 16, FP8 8.5, 4-bit 4.85. “Fits” = total ≤ 90% of memory; Mac GPU can use ~2/3 of unified memory (≤36GB) or ~3/4 (larger). For MoE, all experts must be in memory, so size by total parameters. Method →
03How to run it
Only runtimes the publisher or the runtime’s official library names for this model.
- vLLM — mentioned in the official model card
- SGLang — mentioned in the official model card
- Transformers — mentioned in the official model card
04Official repositories & quantised builds
2 official repo(s) from Qwen.
05Other Qwen models
All 112 →| Model | Params | 4-bit total | Released |
|---|---|---|---|
| Qwen3.8-27B · same series | 27.8B | 19.6 GB | 2026-08-05 |
| Qwen-Image-2.1-Turbo | 7.12B | 5.2 GB | 2026-10-09 |
| Qwen-Image-2.1 | 7.12B | 5.2 GB | 2026-09-14 |
| Qwen3.8-Flash-Next | 180B / 6B | 121 GB | 2026-08-24 |
| Qwen3.6-27B | 27.8B | 19.6 GB | 2026-04-21 |
| Qwen3.6-35B-A3B | 36B / 3B | 24.6 GB | 2026-04-15 |
| Qwen3.5-0.8B | 873M | 1.2 GB | 2026-02-28 |
| Qwen3.5-2B | 2.27B | 2.1 GB | 2026-02-28 |
| Qwen3.5-4B | 4.66B | 3.9 GB | 2026-02-27 |
06Sources
Data checked 2026-08-12. Parameters, context and licence are copied from the official repository; memory figures are our estimates.