LLM Price Index — August 2026
GPT-5 launches below GPT-4 Turbo pricing — industry average input cost falls 18% month-over-month
August 2026 marks the most significant month for LLM pricing since Claude 3 Haiku launched in early 2024. OpenAI's GPT-5 family launched at prices well below initial analyst expectations: GPT-5.4 at $2.50/M input (vs GPT-4o at $5.00/M), and GPT-5 mini at $0.25/M — the cheapest OpenAI model ever. Anthropic's Claude pricing held steady. Google's Gemini 2.5 Flash dropped 40% vs 1.5 Flash. The net effect: industry average input cost (weighted by market share) fell approximately 18% month-over-month.
Biggest movers this month
Cheapest model per capability tier
| Tier | Cheapest model | Price |
|---|---|---|
| Budget (<$0.30/M input) | Gemini 2.5 Flash | $0.075/M |
| Mid-range ($0.30–$3/M input) | DeepSeek V3 | $0.27/M |
| Premium (>$3/M input) | GPT-5.4 | $2.50/M |
Analysis
The GPT-5 launch below GPT-4 pricing represents a structural shift: OpenAI is optimizing for volume and ecosystem lock-in over margin. At $0.25/M, GPT-5 mini directly competes with every open-weight inference provider. DeepSeek V3 at $0.27/M remains the most compelling value in the mid-tier — near-GPT-5.4 quality at near-GPT-5-mini price.
Frequently Asked Questions
Where can I see historical LLM prices?
The WiseCrewAI changelog tracks every price change with date and source URL. Every model page includes a price history table.
WiseCrewAI LLM Price Index publishes monthly. See methodology · sources.