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.
Compare cost per million records processed via DeepSeek V4.1 Flash (V3 successor) vs. Western frontier models.
Multi-turn history expands average input context to 2,000 tokens/req. Output accounts for 71% of total spend.
| Model | Delta | |||
|---|---|---|---|---|
| GPT-4.1 nano (Ultra-Low Cost)OpenAI | $0.10 | $0.40 | $188.33 | Save $80 |
| Claude Haiku 5.5 (High-Volume)Anthropic | $0.10 | $0.50 | $214.13 | Save $54 |
| DeepSeek V4.1 Flash (Off-Peak)DeepSeek | $0.15 | $0.60 | $268.32 | Active Baseline |
| GPT-4o mini (Legacy Budget)OpenAI | $0.15 | $0.60 | $297.91 | +$30 |
| GPT-4.1 mini (Balanced Fast)OpenAI | $0.40 | $1.60 | $753.33 | +$485 |
| Llama 3.3 / 4 70B-Class (Hosted API)Open-Weight Host | $0.59 | $0.79 | $791.22 | +$523 |
| Gemini 3.5 Flash-Lite (Budget Edge)Google | $0.30 | $2.50 | $961.98 | +$694 |
| DeepSeek V4 Pro (Off-Peak Reasoning)DeepSeek | $0.66 | $1.98 | $973.3 | +$705 |
| Gemini 3.8 Flash (Intro Paid Tier)Google | $0.75 | $3.75 | $1,605.96 | +$1,338 |
| o3-mini (Reasoning)OpenAI | $1.10 | $4.40 | $2,184.65 | +$1,916 |
| Gemini 4 HydrogenGoogle | $1.50 | $7.50 | $3,181.1 | +$2,913 |
| GPT-4.1 (1M Long-Context)OpenAI | $2.00 | $8.00 | $3,766.64 | +$3,498 |
| GPT-5 (Flagship Reasoning)OpenAI | $1.25 | $10.00 | $3,875.07 | +$3,607 |
| Claude Sonnet 5.5 (Coding & Agents)Anthropic | $2.00 | $10.00 | $4,241.47 | +$3,973 |
| GPT-6.2 SolOpenAI | $2.00 | $10.00 | $4,241.47 | +$3,973 |
| Gemini 4 CarbonGoogle | $2.00 | $10.00 | $4,241.47 | +$3,973 |
| Gemini 3.1 Pro (Long-Context)Google | $2.00 | $12.00 | $4,921.74 | +$4,653 |
| GPT-4o (Legacy Omni)OpenAI | $2.50 | $10.00 | $4,965.12 | +$4,697 |
| Gemini 4 NeonGoogle | $3.00 | $15.00 | $6,362.2 | +$6,094 |
| Claude Opus 5.5 (Agentic Frontier)Anthropic | $4.00 | $20.00 | $8,482.93 | +$8,215 |
| Claude Fable 5.2Anthropic | $8.00 | $40.00 | $16,883.68 | +$16,615 |
| GPT-6.2 AstraOpenAI | $8.00 | $40.00 | $16,965.86 | +$16,698 |
Ideal baseline for comparing hosted API pricing against self-hosted GPUs. Pre-populated at 15,000 daily requests with 2,000 input tokens, 700 output tokens, 45% prompt caching, and 1 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.