Cloud vs. Rig TCO
Plot cumulative 36-month API spending against hourly RunPod/H100 rentals and owned RTX/Mac hardware with electricity and resale value.
Open Workbench
Why Gemma’s 16 KV heads expand long-context memory faster than Qwen 32B.
OLLAMA_FLASH_ATTENTION=1 OLLAMA_KV_CACHE_TYPE=q8_0 \ ollama run gemma2:27b-instruct-q4_k_m --ctx-size 8192
| Quantization | Memory Needed | Speed Est. | Placement |
|---|---|---|---|
| FP16 / BF16 (16-bit)Reference Lossless | 56.74 GB | 1.2 tok/s | RAM Offload |
| Q8_0 (8-bit)Near-Lossless | 31.24 GB | 3.3 tok/s | RAM Offload |
| Q6_K (6-bit K-Quant)Studio Fidelity | 24.64 GB | 5 tok/s | RAM Offload |
| Q5_K_M (5-bit Medium)Minimal Loss | 21.72 GB | 33.3 tok/s | Pure GPU VRAM |
| Q4_K_M (4-bit Sweet Spot)Recommended Default | 18.83 GB | 43.4 tok/s | Pure GPU VRAM |
| IQ3_XS (3.3-bit I-Quant)Noticeable Degradation | 13.9 GB | 61.6 tok/s | Pure GPU VRAM |
| IQ2_XXS (2.1-bit Extreme)Emergency Fit Only | 9.82 GB | 94.3 tok/s | Pure GPU VRAM |
| Model | Best Pure-VRAM Quant | Memory | Est. Speed |
|---|---|---|---|
| Llama 3.1 8B InstructUltra-fast RAG extraction, IDE autocomplete & classification | FP16 | 17.46 GB | 45 tok/s |
| Qwen 2.5 Coder 14BMid-tier local VS Code copilot balancing speed & repo accuracy | Q8_0 | 17.27 GB | 46.1 tok/s |
| Mistral Small 24BClean instruction following on 24 GB single-GPU workstations | Q6_K | 21.21 GB | 33 tok/s |
| Gemma 2 / 3 27BMultilingual & editorial writing in single 24–32 GB GPUs | Q5_K_M | 21.72 GB | 33.3 tok/s |
| Qwen 2.5 32B / DeepSeek-R1 32BGold-standard local coding & reasoning on RTX 3090/4090/5090 & 48GB Macs | Q4_K_M | 21.6 GB | 32.9 tok/s |
| Mixtral 8x7B MoE (46.7B Total)Sparse Mixture-of-Experts: needs 47B RAM footprint, computes at 13B speed | IQ3_XS | 21.55 GB | 117.5 tok/s |
| Llama 3.3 70B / DeepSeek-R1 70BGPT-4 class private local reasoning for 48 GB–128 GB rigs | IQ2_XXS | 21.87 GB | 33.3 tok/s |
| DeepSeek R1 671B MoE (Full Frontier)Cluster or Mac Studio Ultra / dual EPYC server deployment (37B active per token) | Needs Offload / Cluster | 408.95 GB | OOM |
| Qwen3-30B-A3B (MoE — 3.3B Active)Ultra-fast 30B Sparse MoE decoding at 3.3B speed; effortless fit on 24 GB cards | Q4_K_M | 19.77 GB | 352.8 tok/s |
| Qwen3-32B (Dense Dual-Thinking)2026 dense local reasoning & coding benchmark with dual thinking/non-thinking toggle | Q4_K_M | 21.78 GB | 32.6 tok/s |
| Llama 4 Scout 109B (MoE — 17B Active)Multimodal 109B MoE (17B active) built for 64–128 GB Macs, Dual-3090, or H100 pods | Needs Offload / Cluster | 68.03 GB | 6 tok/s |
Pre-configured telemetry for Gemma 2 / 3 27B (Q4_K_M) on NVIDIA RTX 4090 (24 GB). Requires 18.83 GB (16.49 GB weights + 1.44 GB Q8 KV cache) against 22.8 GB usable pool for ~43.4 tok/s decode.
Primary Technical & US Grid Benchmarks
Plot cumulative 36-month API spending against hourly RunPod/H100 rentals and owned RTX/Mac hardware with electricity and resale value.
Calculate daily, monthly, and annual kWh bills using 80+ PSU efficiency, duty cycle, optional AC cooling load, and 50 US state EIA rates.
Combine dual-GPU draw, CPU peak boost, NVMe arrays, and ATX 3.1 transient spikes to hit the 50–70% PSU efficiency sweet spot.