MODEL ARCHITECTURE SPEC • MISTRAL AI123B Dense
Mistral Large 2 (123B Instruct) VRAM & inference speed
123B Total ParamsDense Architecture88 Layers • 8 KV Heads32K Default Ctx (Max 128K)Ollama: mistral-large:123b
Accurate memory requirements, KV cache footprint, and tokens/sec generation benchmarks for running Mistral Large 2 (123B Instruct) across local GPUs, Apple Silicon Macs, and cloud instances.
✓ FITS IN VRAM83.04 GB (75%)~2 t/s
Mistral Large 2 (123B Instruct) on AMD Strix Halo Mini-PC (Ryzen AI Max+ 395 128GB)
✓ Fits Comfortably in VRAM
✓ FITS IN VRAM
Required VRAM
83.04 GB
75% of hardware capacity (110 GB)Est. Decode Speed
~2 tok/s
Dense model @ 256 GB/sAvailable Headroom
+26.96 GB
Safe for OS display & full contextMemory Allocation & Headroom
83.04 GB allocated of 110 GB capacity
75% utilized
Weights: 73.8 GB
KV: (32K): 5.91 GB
Overhead: 3.33 GB
Free Headroom: +26.96 GB
⚡ Architecture & Throughput Transparency: Single-user decode speed (~2 tok/s) is memory-bandwidth bound at batch size = 1 on AMD Strix Halo Mini-PC (Ryzen AI Max+ 395 128GB) (256 GB/s). Prompt prefill / TTFT (~29 tok/s) is compute-bound and utilizes GPU Tensor cores. Actual latency varies with FlashAttention-2 and runtime engine (Ollama/vLLM/llama.cpp).
OLLAMA_FLASH_ATTENTION=1 OLLAMA_KV_CACHE_TYPE=q8_0 ollama run mistral-large:123bCloud Instance Options for Mistral Large 2 (123B Instruct)
On-DemandDeploy this model on an on-demand cloud GPU without local hardware constraints or power usage:
RunPod Cloud GPUsBest 1-Click Ollama & vLLM
$0.34/hr (RTX 4070 Ti) – $2.39/hr (H100)Vast.ai GPU MarketplaceCheapest Dual RTX 3090 / 4090
$0.18/hr (RTX 3060) – $0.55/hr (RTX 4090)Lambda GPU CloudEnterprise 8x H100 Clusters
$1.99/hr (GH200 / H100)Quantization Precision Matrix
32K ctx • KV: Q8_0| Quantization | Weights | KV Cache | Total VRAM | Est. Speed | Fit Status |
|---|---|---|---|---|---|
| FP1616 bpw | 246 GB | 5.91 GB | 261.27 GB (238%) | 0 t/s | OOM |
| Q8_08.5 bpw | 130.38 GB | 5.91 GB | 141.6 GB (129%) | 0 t/s | PARTIAL |
| Q6_K6.56 bpw | 100.86 GB | 5.91 GB | 111.05 GB (101%) | 0 t/s | PARTIAL |
| Q5_K_M5.68 bpw | 87.33 GB | 5.91 GB | 97.05 GB (88%) | ~2 t/s | FITS |
| Q4_K_M4.8 bpw | 73.8 GB | 5.91 GB | 83.04 GB (75%) | ~2 t/s | FITS |
| FP44.25 bpw | 65.19 GB | 5.91 GB | 74.13 GB (67%) | ~2 t/s | FITS |
| IQ3_XS3.3 bpw | 50.43 GB | 5.91 GB | 58.86 GB (54%) | ~3 t/s | FITS |
Export Full Architecture & Infrastructure Spec
Get an executive-ready PDF report with exact memory math, hardware benchmarks, and cost comparisons.
Hardware Compatibility Matrix for Mistral Large 2 (123B Instruct)
Standard Q4_K_M precision with 32K context window and Q8_0 KV cache
| Hardware System | VRAM Envelope | Required VRAM | Headroom | Est. Decode Speed | Compatibility | Test Fit |
|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 5090 (32GB GDDR7) Consumer GPU • 1792 GB/s | 32 GB | 83.04 GB | -51.04 GB | 0 (OOM) | ✕ OOM | Test Fit → |
| NVIDIA GeForce RTX 4090 (24GB GDDR6X) Consumer GPU • 1008 GB/s | 24 GB | 83.04 GB | -59.04 GB | 0 (OOM) | ✕ OOM | Test Fit → |
| 2x NVIDIA RTX 3090 Dual Rig (48GB Total VRAM) Multi-GPU Rig • 936 GB/s | 48 GB | 83.04 GB | -35.04 GB | 0 (OOM) | ✕ OOM | Test Fit → |
| NVIDIA GeForce RTX 5080 (16GB GDDR7) Consumer GPU • 960 GB/s | 16 GB | 83.04 GB | -67.04 GB | 0 (OOM) | ✕ OOM | Test Fit → |
| NVIDIA RTX 4070 Ti Super (16GB) Consumer GPU • 672 GB/s | 16 GB | 83.04 GB | -67.04 GB | 0 (OOM) | ✕ OOM | Test Fit → |
| NVIDIA GeForce RTX 3060 (12GB) Consumer GPU • 360 GB/s | 12 GB | 83.04 GB | -71.04 GB | 0 (OOM) | ✕ OOM | Test Fit → |
| Apple Mac Mini M4 Pro (64GB Unified Memory) Apple Silicon • 273 GB/s | 54 GB | 83.04 GB | -29.04 GB | 0 (OOM) | ✕ OOM | Test Fit → |
| Apple Mac Studio M4 Max (128GB Unified Memory) Apple Silicon • 546 GB/s | 112 GB | 83.04 GB | +28.96 GB | ~5 tok/s | ✓ FITS | Test Fit → |
| Apple Mac Studio M5 Ultra (256GB Unified Memory) Apple Silicon • 890 GB/s | 236 GB | 83.04 GB | +152.96 GB | ~8 tok/s | ✓ FITS | Test Fit → |
| AMD Strix Halo Mini-PC (Ryzen AI Max+ 395 128GB) Unified APU / Mini-PC • 256 GB/s | 110 GB | 83.04 GB | +26.96 GB | ~2 tok/s | ✓ FITS | Test Fit → |
| NVIDIA DGX Spark (128GB Grace-Blackwell Unified) AI Appliance • 273 GB/s | 118 GB | 83.04 GB | +34.96 GB | ~2 tok/s | ✓ FITS | Test Fit → |
| NVIDIA H100 SXM (80GB HBM3 Cloud GPU) Datacenter Cloud GPU • 3350 GB/s | 80 GB | 83.04 GB | -3.04 GB | ~7 tok/s | ⚡ PARTIAL | Test Fit → |
| NVIDIA GeForce RTX 5070 Ti (16GB GDDR7) Consumer GPU • 896 GB/s | 16 GB | 83.04 GB | -67.04 GB | 0 (OOM) | ✕ OOM | Test Fit → |
| NVIDIA GeForce RTX 5070 (12GB GDDR7) Consumer GPU • 672 GB/s | 12 GB | 83.04 GB | -71.04 GB | 0 (OOM) | ✕ OOM | Test Fit → |
| NVIDIA GeForce RTX 4080 Super (16GB GDDR6X) Consumer GPU • 736 GB/s | 16 GB | 83.04 GB | -67.04 GB | 0 (OOM) | ✕ OOM | Test Fit → |
| AMD Radeon RX 7900 XTX (24GB GDDR6) Consumer GPU • 960 GB/s | 24 GB | 83.04 GB | -59.04 GB | 0 (OOM) | ✕ OOM | Test Fit → |
| AMD Radeon RX 7900 XT (20GB GDDR6) Consumer GPU • 800 GB/s | 20 GB | 83.04 GB | -63.04 GB | 0 (OOM) | ✕ OOM | Test Fit → |
| Apple Mac Studio M2 Ultra (192GB Unified Memory) Apple Silicon Workstation • 800 GB/s | 144 GB | 83.04 GB | +60.96 GB | ~7 tok/s | ✓ FITS | Test Fit → |