Mistral-Nemo-Instruct-2407 on RTX 4070 Ti SUPER: Fits (Sep 2026)
Conditional memory judgement for one configuration. Change context, quant, or engine and the conclusion can change. Configuration used on this page: GGUF Q6_K (Unsloth), 8,192 cached tokens, 1 sequence, engine llama.cpp. Status: Likely fits. Software: Not verified..
Verdict on GeForce RTX 4070 Ti SUPER 16GB (desktop)
Yes, run Q6_K. The calculator puts it at 10.7 GiB–11.4 GiB at 8,192 tokens, inside 15.0 GiB of usable VRAM (after the driver reserve). Q8_0 is the next step up and does not clear the bar (13.5 GiB–14.2 GiB).
With Q6_K, the longest context that still counts as a fit is about 21,113 tokens of a 131,072-token native window, one sequence, fp16 KV cache.
| GGUF quant | File weights | KV cache | Estimated total | Verdict |
|---|---|---|---|---|
| Q8_0 | 12.1 GiB | 1.3 GiB | 13.5 GiB–14.2 GiB | Tight |
| Q6_K (recommended) | 9.4 GiB | 1.3 GiB | 10.7 GiB–11.4 GiB | Fits |
| Q5_K_M | 8.1 GiB | 1.3 GiB | 9.5 GiB–10.1 GiB | Comfortable |
| Q4_K_M | 7.0 GiB | 1.3 GiB | 8.3 GiB–8.9 GiB | Comfortable |
Engine llama.cpp, one sequence, fp16 KV. File weights are the published GGUF sizes in the catalog. Totals come from the same calculator that powers the tool below; they are estimates, not a measurement of this setup.
Fixed page configuration
GGUF Q6_K needs 10.7 GiB-11.4 GiB at 8,192 tokens against 15.00 GiB net on GeForce RTX 4070 Ti SUPER 16GB (desktop). Q8_0 is tight or over budget.
- ModelMistral-Nemo-Instruct-2407
- HardwareGeForce RTX 4070 Ti SUPER 16GB (desktop)
- QuantizationGGUF Q6_K (Unsloth)
- Context length (cached tokens)8,192
- Engine
llama.cpp - Concurrent sequences1
- KV cache typefp16
Technical identifiers and share data
- Model
mistral-nemo-instruct-2407 - Hardware
nvidia-geforce-rtx-4070-ti-super-16gb-desktop - Quantization
gguf-q6_k - Share fragment for this identity
v2.eyJzY2hlbWFWZXJzaW9uIjoyLCJjYWxjdWxhdG9yVmVyc2lvbiI6IjAuMy4wIiwiZGF0YXNldFZlcnNpb24iOiIyMDI2LTEwLTAyLjEiLCJjb25maWciOnsibW9kZWxJZCI6Im1pc3RyYWwtbmVtby1pbnN0cnVjdC0yNDA3IiwiaGFyZHdhcmVJZCI6Im52aWRpYS1nZWZvcmNlLXJ0eC00MDcwLXRpLXN1cGVyLTE2Z2ItZGVza3RvcCIsImRldmljZUNvdW50IjoxLCJxdWFudGl6YXRpb25JZCI6ImdndWYtcTZfayIsImNvbnRleHRUb2tlbnMiOjgxOTIsImNvbmN1cnJlbnRTZXF1ZW5jZXMiOjEsImVuZ2luZSI6ImxsYW1hLmNwcCIsImt2VHlwZSI6ImZwMTYifX0
Memory breakdown
- Weight artifact
- 9.37 GiB (10,056,209,952 bytes)
- Raw KV cache
- 1.25 GiB (1,342,177,280 bytes)
- Engine extras / workspace
- 0.13 GiB–0.77 GiB
- Total estimated demand
- 10.74 GiB–11.39 GiB
- Net calculator budget
- 15.00 GiB
- OS / display reserve
- 1.00 GiB
- Remaining vs net budget
- 3.61 GiB–4.26 GiB
11.39 GiB high estimate vs 15.00 GiB net
Enough memory does not mean this engine, quant, and OS will run. Support does not mean it will be fast.
Other published files on this device
Same model, device and context (8,192 tokens, 1 sequence, FP16 KV). Each row is one published weight-file set with its usual engine, sorted by memory.
| Quantization | Engine | Estimated memory | Fit |
|---|---|---|---|
| BF16 safetensors | vllm | 24.69 GiB–26.60 GiB | Does not fit |
| GGUF Q8_0 (Unsloth) | llama.cpp | 13.50 GiB–14.21 GiB | Tight fit |
| GGUF Q6_K (Unsloth) This page | llama.cpp | 10.74 GiB–11.39 GiB | Fits |
| GGUF Q5_K_M (Unsloth) | llama.cpp | 9.50 GiB–10.13 GiB | Runs comfortably |
| GGUF Q4_K_M (Unsloth) | llama.cpp | 8.34 GiB–8.94 GiB | Runs comfortably |
Assumptions that affect this result
- Requested context is within the recorded native window of 131072 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-Nemo-Instruct-2407.Q6_K.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.
Related pages
Sources
- Mistral-Nemo-Instruct-2407 model card
- config.json — revision 04d8a90549d23fc6bd7f642064003592df51e9b3
- safetensors index — metadata.total_size=24495564800
- Artifact repo unsloth/Mistral-Nemo-Instruct-2407-GGUF
- NVIDIA GeForce RTX 40 Series specs
Calculator
This combo page keeps the route model and device. Customize to change them on the main calculator; the share then keeps that identity.
Loading catalog snapshot…
Search or pick a catalog checkpoint to see memory demand.