Qwen3-8B-Base VRAM & Memory Requirements
Specific checkpoints with sourced architecture and weight files. GGUF Q4_K_M (mradermacher): 5.93 GiB–6.48 GiB at 8,192 tokens, 1 sequence, FP16 KV, llama.cpp.
This result is an estimate, not a measured trace.Not verifiedFile-backed weights
Runs on
Publisher and provenance
- PublisherQwen
- Hugging Face
Qwen/Qwen3-8B-Base - Revision
49e3418fbbbca6ecbdf9608b4d22e5a407081db4 - Total parameters8,190,735,360
- Layers36
- Query heads32
- KV heads8 (gqa)
- Head dimension128
- Recorded context capability32,768
- Completenesscomplete · full
Memory layers in this estimate
Artifact bytes, raw KV, engine extras, and total demand are separate. This is not a runtime trace.
- Weight artifact
- 4.68 GiB (5,027,783,936 bytes)
- Raw KV cache
- 1.13 GiB (1,207,959,552 bytes)
- Engine extras / workspace
- 0.13 GiB–0.67 GiB
- Total estimated demand
- 5.93 GiB–6.48 GiB
Published artifacts
| Quantization | Selected files | Bytes | Engine | Evidence |
|---|---|---|---|---|
| BF16 safetensors bf16 | 16,383,567,872 (15.26 GiB / 16.38 GB) | vllm, sglang, unspecified | file-metadata | |
| GGUF Q4_K_M (mradermacher) gguf-q4_k_m | Qwen3-8B-Base.Q4_K_M.gguf | 5,027,783,936 (4.68 GiB / 5.03 GB) | llama.cpp, ollama, unspecified | file-metadata |
| GGUF Q5_K_M (mradermacher) gguf-q5_k_m | Qwen3-8B-Base.Q5_K_M.gguf | 5,851,112,704 (5.45 GiB / 5.85 GB) | llama.cpp, ollama, unspecified | file-metadata |
| GGUF Q6_K (mradermacher) gguf-q6_k | Qwen3-8B-Base.Q6_K.gguf | 6,725,899,520 (6.26 GiB / 6.73 GB) | llama.cpp, ollama, unspecified | file-metadata |
| GGUF Q8_0 (mradermacher) gguf-q8_0 | Qwen3-8B-Base.Q8_0.gguf | 8,709,518,592 (8.11 GiB / 8.71 GB) | llama.cpp, ollama, unspecified | file-metadata |
| INT4 theoretical | No published file | 4,095,367,680 (theoretical) | — | theoretical |
Memory by quantization and context
Concurrent sequences = 1. KV defaults to FP16. The range includes documented runtime extras for the listed engine. These figures use the same core as the interactive calculator.
| Quantization | 2,048 tokens | 8,192 tokens | 32,768 tokens |
|---|---|---|---|
| BF16 safetensors | 16.16 GiB–18.05 GiB | 17.01 GiB–18.92 GiB | 20.38 GiB–22.39 GiB |
| GGUF Q4_K_M (mradermacher) | 5.09 GiB–5.56 GiB | 5.93 GiB–6.48 GiB | 9.31 GiB–10.09 GiB |
| GGUF Q5_K_M (mradermacher) | 5.86 GiB–6.35 GiB | 6.70 GiB–7.26 GiB | 10.07 GiB–10.87 GiB |
| GGUF Q6_K (mradermacher) | 6.67 GiB–7.18 GiB | 7.51 GiB–8.09 GiB | 10.89 GiB–11.70 GiB |
| GGUF Q8_0 (mradermacher) | 8.52 GiB–9.06 GiB | 9.36 GiB–9.98 GiB | 12.74 GiB–13.59 GiB |
| INT4 theoretical Theoretical rows are not downloadable artifacts. | 4.22 GiB–5.82 GiB | 5.06 GiB–6.87 GiB | 8.44 GiB–11.02 GiB |
Raw KV contribution
Standard GQA/MHA raw KV is 2 × layers × kvHeads × headDim × tokens × sequences × bytes per element. This page uses KV heads (8), not query heads (32). Using query heads would overstate this cache by 4×.
- 8,192 tokens, 1 sequence, FP16: 1,207,959,552 bytes (1.1250 GiB)
- 32,768 tokens: 4,831,838,208 bytes (4.5000 GiB)
Related pages
Sources
- Qwen3-8B-Base model card
- config.json — revision 49e3418fbbbca6ecbdf9608b4d22e5a407081db4
- safetensors index — metadata.total_size=16383567872
- Artifact repo mradermacher/Qwen3-8B-Base-GGUF
Assumptions that affect this result
- Requested context is within the recorded native window of 32768 tokens.
- Full-resident weight estimates use total parameters, including inactive MoE experts.
- Weight bytes come from the selected weight files only, not from every file in the repository.
- Download file size is not peak GPU memory. Runtime layout, allocator padding, and KV are extra.
- Counted files: Qwen3-8B-Base.Q4_K_M.gguf.
- Context budget is the number of cached tokens per sequence (prompt plus reserved generation).
- Concurrent sequences share one model replica; weights are not multiplied by sequence count.
- KV dtype is independent of weight quantization unless you change it.
Calculator
Loading catalog snapshot…
Search or pick a catalog checkpoint to see memory demand.
Search or pick a catalog checkpoint to see memory demand.