Qwen2.5-Coder-3B-Instruct VRAM & Memory Requirements
Specific checkpoints with sourced architecture and weight files. GGUF Q4_K_M (Qwen): 2.37 GiB–2.80 GiB at 8,192 tokens, 1 sequence, FP16 KV, llama.cpp.
This result is an estimate, not a measured trace.Not verifiedFile-backed weights
Publisher and provenance
- PublisherQwen
- Hugging Face
Qwen/Qwen2.5-Coder-3B-Instruct - Revision
488639f1ff808d1d3d0ba301aef8c11461451ec5 - Total parameters3,085,938,688
- Layers36
- Query heads16
- KV heads2 (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
- 1.96 GiB (2,104,932,800 bytes)
- Raw KV cache
- 0.28 GiB (301,989,888 bytes)
- Engine extras / workspace
- 0.13 GiB–0.56 GiB
- Total estimated demand
- 2.37 GiB–2.80 GiB
Published artifacts
| Quantization | Selected files | Bytes | Engine | Evidence |
|---|---|---|---|---|
| BF16 safetensors bf16 | model-00001-of-00002.safetensors, model-00002-of-00002.safetensors | 6,171,877,376 (5.75 GiB / 6.17 GB) | vllm, sglang, unspecified | file-metadata |
| GGUF Q4_K_M (Qwen) gguf-q4_k_m | qwen2.5-coder-3b-instruct-q4_k_m.gguf | 2,104,932,800 (1.96 GiB / 2.10 GB) | llama.cpp, ollama, unspecified | file-metadata |
| GGUF Q5_K_M (Qwen) gguf-q5_k_m | qwen2.5-coder-3b-instruct-q5_k_m.gguf | 2,438,740,416 (2.27 GiB / 2.44 GB) | llama.cpp, ollama, unspecified | file-metadata |
| GGUF Q6_K (Qwen) gguf-q6_k | qwen2.5-coder-3b-instruct-q6_k.gguf | 2,793,411,008 (2.60 GiB / 2.79 GB) | llama.cpp, ollama, unspecified | file-metadata |
| GGUF Q8_0 (Qwen) gguf-q8_0 | qwen2.5-coder-3b-instruct-q8_0.gguf | 3,616,088,512 (3.37 GiB / 3.62 GB) | llama.cpp, ollama, unspecified | file-metadata |
| INT4 theoretical | No published file | 1,542,969,344 (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 | 6.44 GiB–8.32 GiB | 6.65 GiB–8.54 GiB | 7.50 GiB–9.41 GiB |
| GGUF Q4_K_M (Qwen) | 2.16 GiB–2.57 GiB | 2.37 GiB–2.80 GiB | 3.21 GiB–3.70 GiB |
| GGUF Q5_K_M (Qwen) | 2.47 GiB–2.89 GiB | 2.68 GiB–3.12 GiB | 3.52 GiB–4.02 GiB |
| GGUF Q6_K (Qwen) | 2.80 GiB–3.23 GiB | 3.01 GiB–3.45 GiB | 3.85 GiB–4.36 GiB |
| GGUF Q8_0 (Qwen) | 3.56 GiB–4.01 GiB | 3.77 GiB–4.24 GiB | 4.62 GiB–5.14 GiB |
| INT4 theoretical Theoretical rows are not downloadable artifacts. | 1.63 GiB–2.51 GiB | 1.84 GiB–2.77 GiB | 2.69 GiB–3.81 GiB |
Raw KV contribution
Standard GQA/MHA raw KV is 2 × layers × kvHeads × headDim × tokens × sequences × bytes per element. This page uses KV heads (2), not query heads (16). Using query heads would overstate this cache by 8×.
- 8,192 tokens, 1 sequence, FP16: 301,989,888 bytes (0.2813 GiB)
- 32,768 tokens: 1,207,959,552 bytes (1.1250 GiB)
Related pages
Sources
- Qwen2.5-Coder-3B-Instruct model card
- config.json — revision 488639f1ff808d1d3d0ba301aef8c11461451ec5
- safetensors index — metadata.total_size=6171877376
- Artifact repo Qwen/Qwen2.5-Coder-3B-Instruct-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: qwen2.5-coder-3b-instruct-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.