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.
Stress-test Claude Opus 5.5 ($20/1M output standard, $40/1M Fast Mode) on legal/financial audit pipelines before shipping.
Multi-turn history expands average input context to 19,090 tokens/req. Output accounts for 50% of total spend.
| Model | Delta | |||
|---|---|---|---|---|
| GPT-4.1 nano (Ultra-Low Cost)OpenAI | $0.10 | $0.40 | $24.27 | Save $951 |
| Claude Haiku 5.5 (High-Volume)Anthropic | $0.10 | $0.50 | $24.96 | Save $950 |
| DeepSeek V4.1 Flash (Off-Peak)DeepSeek | $0.15 | $0.60 | $32.39 | Save $943 |
| GPT-4o mini (Legacy Budget)OpenAI | $0.15 | $0.60 | $40.76 | Save $934 |
| GPT-4.1 mini (Balanced Fast)OpenAI | $0.40 | $1.60 | $97.07 | Save $878 |
| Gemini 3.5 Flash-Lite (Budget Edge)Google | $0.30 | $2.50 | $99.22 | Save $876 |
| DeepSeek V4 Pro (Off-Peak Reasoning)DeepSeek | $0.66 | $1.98 | $127.47 | Save $848 |
| Llama 3.3 / 4 70B-Class (Hosted API)Open-Weight Host | $0.59 | $0.79 | $156.36 | Save $819 |
| Gemini 3.8 Flash (Intro Paid Tier)Google | $0.75 | $3.75 | $187.19 | Save $788 |
| o3-mini (Reasoning)OpenAI | $1.10 | $4.40 | $298.89 | Save $676 |
| Gemini 4 HydrogenGoogle | $1.50 | $7.50 | $365.66 | Save $609 |
| GPT-5 (Flagship Reasoning)OpenAI | $1.25 | $10.00 | $403.29 | Save $572 |
| GPT-4.1 (1M Long-Context)OpenAI | $2.00 | $8.00 | $485.33 | Save $490 |
| Claude Sonnet 5.5 (Coding & Agents)Anthropic | $2.00 | $10.00 | $487.54 | Save $488 |
| GPT-6.2 SolOpenAI | $2.00 | $10.00 | $487.54 | Save $488 |
| Gemini 4 CarbonGoogle | $2.00 | $10.00 | $487.54 | Save $488 |
| Gemini 3.1 Pro (Long-Context)Google | $2.00 | $12.00 | $547.86 | Save $427 |
| GPT-4o (Legacy Omni)OpenAI | $2.50 | $10.00 | $679.29 | Save $296 |
| Gemini 4 NeonGoogle | $3.00 | $15.00 | $731.31 | Save $244 |
| Claude Opus 5.5 (Agentic Frontier)Anthropic | $4.00 | $20.00 | $975.08 | Active Baseline |
| Claude Fable 5.2Anthropic | $8.00 | $40.00 | $1,926.93 | +$952 |
| GPT-6.2 AstraOpenAI | $8.00 | $40.00 | $1,950.17 | +$975 |
Comparing frontier per-query spend helps identify when batch or routing is mandatory. Pre-populated at 400 daily requests with 18,000 input tokens, 2,000 output tokens, 50% prompt caching, and 2 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.