Mistral-Small-24B-Instruct-2501 VRAM & Memory Requirements
Specific checkpoints with sourced architecture and weight files. GGUF Q4_K_M (Unsloth): 14.72 GiB–15.45 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
- Can Mistral-Small-24B-Instruct-2501 run on RTX 4090?
- Can Mistral-Small-24B-Instruct-2501 run on RTX 3090?
- Can Mistral-Small-24B-Instruct-2501 run on RTX 3060?
- Can Mistral-Small-24B-Instruct-2501 run on RTX 4060 Ti?
- Can Mistral-Small-24B-Instruct-2501 run on RTX 4070 Ti SUPER?
- Can Mistral-Small-24B-Instruct-2501 run on RX 7900 XTX?
Publisher and provenance
- PublisherMistral AI
- Hugging Face
mistralai/Mistral-Small-24B-Instruct-2501 - Revision
9527884be6e5616bdd54de542f9ae13384489724 - Total parameters23,572,403,200
- Layers40
- 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
- 13.35 GiB (14,333,909,248 bytes)
- Raw KV cache
- 1.25 GiB (1,342,177,280 bytes)
- Engine extras / workspace
- 0.13 GiB–0.85 GiB
- Total estimated demand
- 14.72 GiB–15.45 GiB
Published artifacts
| Quantization | Selected files | Bytes | Engine | Evidence |
|---|---|---|---|---|
| BF16 safetensors bf16 | 47,144,806,400 (43.91 GiB / 47.14 GB) | vllm, sglang, unspecified | file-metadata | |
| GGUF Q4_K_M (Unsloth) gguf-q4_k_m | Mistral-Small-24B-Instruct-2501-Q4_K_M.gguf | 14,333,909,248 (13.35 GiB / 14.33 GB) | llama.cpp, ollama, unspecified | file-metadata |
| GGUF Q6_K (Unsloth) gguf-q6_k | Mistral-Small-24B-Instruct-2501-Q6_K.gguf | 19,345,938,688 (18.02 GiB / 19.35 GB) | llama.cpp, ollama, unspecified | file-metadata |
| GGUF Q8_0 (Unsloth) gguf-q8_0 | Mistral-Small-24B-Instruct-2501.Q8_0.gguf | 25,054,779,648 (23.33 GiB / 25.05 GB) | llama.cpp, ollama, unspecified | file-metadata |
| GGUF Q5_K_M (bartowski) gguf-q5_k_m | Mistral-Small-24B-Instruct-2501-Q5_K_M.gguf | 16,763,983,552 (15.61 GiB / 16.76 GB) | llama.cpp, ollama, unspecified | file-metadata |
| INT4 theoretical | No published file | 11,786,201,600 (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 | 44.84 GiB–46.73 GiB | 45.78 GiB–47.69 GiB | 49.53 GiB–51.56 GiB |
| GGUF Q4_K_M (Unsloth) | 13.79 GiB–14.44 GiB | 14.72 GiB–15.45 GiB | 18.47 GiB–19.47 GiB |
| GGUF Q6_K (Unsloth) | 18.45 GiB–19.20 GiB | 19.39 GiB–20.22 GiB | 23.14 GiB–24.23 GiB |
| GGUF Q8_0 (Unsloth) | 23.77 GiB–24.62 GiB | 24.71 GiB–25.64 GiB | 28.46 GiB–29.65 GiB |
| GGUF Q5_K_M (bartowski) | 16.05 GiB–16.74 GiB | 16.99 GiB–17.76 GiB | 20.74 GiB–21.77 GiB |
| INT4 theoretical Theoretical rows are not downloadable artifacts. | 11.41 GiB–15.08 GiB | 12.35 GiB–16.25 GiB | 16.10 GiB–20.87 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: 1,342,177,280 bytes (1.2500 GiB)
- 32,768 tokens: 5,368,709,120 bytes (5.00 GiB)
Related pages
Sources
- Mistral-Small-24B-Instruct-2501 model card
- config.json — revision 9527884be6e5616bdd54de542f9ae13384489724
- safetensors index — metadata.total_size=47144806400
- Artifact repo unsloth/Mistral-Small-24B-Instruct-2501-GGUF
- Artifact repo bartowski/Mistral-Small-24B-Instruct-2501-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: Mistral-Small-24B-Instruct-2501-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.