API Prices · History
GPT-4-class API price history
The cheapest list price for a model matching original GPT-4, per million tokens, from GPT-4's launch.
Archived benchmark-qualified observations through Apr 2026. The series tracks a changing set of models that met a historical quality threshold; it is separate from the current price comparison.
The price of GPT-4-class intelligence
Historical research · US$ per 1M tokens, fixed 3:1 mix · log scale
More actions for The price of GPT-4-class intelligence
Data — endpoints, license, and freshness behind The price of GPT-4-class intelligence
The current LLM API price table in USD per million tokens: input, output, cache-hit and batch prices where recorded, the long-context tier, and the blended 3:1 price per model, with each row's review date, terms and source — plus the dated GPT-4-class price-floor history.
curl -s https://aicomputetracker.com/token-prices.jsonNo API key, CORS open (Access-Control-Allow-Origin: *) — fetch it from a browser, a notebook, or a cron job.
The site’s compilation is licensed CC BY 4.0 — attribution for this chart’s data “AI Compute Tracker (aicomputetracker.com), CC BY 4.0; third-party terms apply to data from Model API pricing pages”. Third-party source data — such as Kalshi market data and providers’ list prices — remains subject to its sources’ terms.
Core data · re-survey due
Cite — copy a citation for The price of GPT-4-class intelligence
Copy-ready citation
Cite as: AI Compute Tracker, “The price of GPT-4-class intelligence”, core data as of 2026-07-23, https://aicomputetracker.com/ai-model-api-prices/history/#token-price-collapse, AI Compute Tracker (aicomputetracker.com), CC BY 4.0; third-party terms apply to data from Model API pricing pages, accessed <date you visit>
Replace “<date you visit>” with your access date.
Source and methodology
Historical observations through Apr 2026; not today’s lowest offer. Official pricing pages & launch announcements. ‘GPT-4-class’ = matches original GPT-4 on MMLU (~86.4).
GPT-4 launched at $37.50 blended. The latest archived benchmark-qualified observation is $0.18 in Apr 2026. Newer table entries have not been independently checked against that quality threshold.
Open a model’s page for its exact named model’s archived observations. The archive records monthly dates without complete billing-tier history, so it does not support reliable 30-day or 90-day model price changes. Missing comparable history stays unknown.
How far did the archived GPT-4-class inference price fall?
In the archived survey through 2026-04, the lowest tracked model matching original-GPT-4 quality was DeepSeek V4 Flash at $0.18 per million tokens blended 3:1 — −99.5% versus GPT-4 at launch ($37.50 blended, Mar 2023). These are historical list prices, not a current lowest-price claim: batch endpoints run about 50% off and cached input about 90% off.