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.
Benchmark daily analyst seat token consumption vs. switching routine steps to GPT-5 or GPT-4.1 mini.
Multi-turn history expands average input context to 6,120 tokens/req. Output accounts for 42% of total spend.
| Model | Delta | |||
|---|---|---|---|---|
| GPT-4.1 nano (Ultra-Low Cost)OpenAI | $0.10 | $0.40 | $61.74 | Save $1,584 |
| Claude Haiku 5.5 (High-Volume)Anthropic | $0.10 | $0.50 | $66.14 | Save $1,579 |
| DeepSeek V4.1 Flash (Off-Peak)DeepSeek | $0.15 | $0.60 | $86.98 | Save $1,558 |
| GPT-4o mini (Legacy Budget)OpenAI | $0.15 | $0.60 | $98.72 | Save $1,547 |
| GPT-4.1 mini (Balanced Fast)OpenAI | $0.40 | $1.60 | $246.95 | Save $1,398 |
| Gemini 3.5 Flash-Lite (Budget Edge)Google | $0.30 | $2.50 | $266.91 | Save $1,378 |
| Llama 3.3 / 4 70B-Class (Hosted API)Open-Weight Host | $0.59 | $0.79 | $328.86 | Save $1,316 |
| DeepSeek V4 Pro (Off-Peak Reasoning)DeepSeek | $0.66 | $1.98 | $338.97 | Save $1,306 |
| Gemini 3.8 Flash (Intro Paid Tier)Google | $0.75 | $3.75 | $496.07 | Save $1,149 |
| o3-mini (Reasoning)OpenAI | $1.10 | $4.40 | $723.95 | Save $921 |
| Gemini 4 HydrogenGoogle | $1.50 | $7.50 | $979.91 | Save $665 |
| GPT-5 (Flagship Reasoning)OpenAI | $1.25 | $10.00 | $1,083.59 | Save $562 |
| GPT-4.1 (1M Long-Context)OpenAI | $2.00 | $8.00 | $1,234.77 | Save $411 |
| Claude Sonnet 5.5 (Coding & Agents)Anthropic | $2.00 | $10.00 | $1,306.54 | Save $339 |
| GPT-6.2 SolOpenAI | $2.00 | $10.00 | $1,306.54 | Save $339 |
| Gemini 4 CarbonGoogle | $2.00 | $10.00 | $1,306.54 | Save $339 |
| Gemini 3.1 Pro (Long-Context)Google | $2.00 | $12.00 | $1,459.81 | Save $186 |
| GPT-4o (Legacy Omni)OpenAI | $2.50 | $10.00 | $1,645.34 | Active Baseline |
| Gemini 4 NeonGoogle | $3.00 | $15.00 | $1,959.82 | +$314 |
| Claude Opus 5.5 (Agentic Frontier)Anthropic | $4.00 | $20.00 | $2,613.09 | +$968 |
| Claude Fable 5.2Anthropic | $8.00 | $40.00 | $5,193.58 | +$3,548 |
| GPT-6.2 AstraOpenAI | $8.00 | $40.00 | $5,226.18 | +$3,581 |
Multi-turn chat replay expands cumulative input tokens quadratically without pruning. Pre-populated at 2,500 daily requests with 4,500 input tokens, 900 output tokens, 35% prompt caching, and 4 turns/session.
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.
Count live prompt tokens across English prose and JSON/code, size RAG document chunks, and compute float32 vs. int8 HNSW vector RAM.
Test Llama 70B, Qwen 32B, or DeepSeek R1 across RTX 3090/4090/5090 & Apple Silicon Macs with GQA KV-cache math and tokens/sec speed.