May 2023
GPT-4 API launches at $30/M input — sets the baseline ceiling for the era.
Nov 2023
GPT-4 Turbo cuts input to $10/M — first significant step down the cost curve.
May 2024
GPT-4o launches at $5/M input; Claude 3 Haiku breaks $1/M output barrier. Competition at the budget tier heats up.
Oct 2024
OpenAI cuts GPT-4o 50% to $2.50/M input. Anthropic drops Claude 3.5 Haiku to $0.80/M. Price wars accelerate across providers.
Q1 2025
Google launches Gemini 2.0 Flash at $0.10/M input — sub-dollar budget tier arrives. xAI and Mistral enter the commercial API market.
Feb 2026
Claude Opus drops 67% — from $15/M to $5/M input. GPT-5 Mini launches at $0.75/M input. Largest single-generation Anthropic price cut on record.
May 2026
Claude Opus 4.8 holds at $5/M. GPT-5.5 sets the new flagship ceiling at $5/$30. Gemini 3.5 Flash launches at $1.50/M — beats 3.1 Pro on coding at 25% lower cost.
Jul 9 2026
GPT-5.6 family (Sol/Terra/Luna) goes GA with 1.05M context. Claude Sonnet 5 launches at introductory $2/M input — Opus-level agentic performance at Sonnet prices. Grok 4.5 drops output to $6/M, the cheapest flagship output tier on the market.
Jul 30 2026
OpenAI cuts GPT-5.6 Terra 20% → $2/M and Luna 80% → $0.20/M. Luna rewrites the budget tier: current-generation quality at near-Nano prices.
Aug 12 2026
Grok 4.6 launches at $2/$6 — 3× output-to-input ratio vs. the industry standard 5–6×. Gemini 3.7 Flash arrives at $0.75/M input, halving the Flash tier. Claude Fable 5 establishes a new premium ceiling at $10/$50.