AI Calculator Pro
Weekly recap

What changed in LLM pricing — Week of 31 August – 6 September 2026

In the week of 31 August – 6 September 2026, we recorded 12 price changes (5 cuts, 7 increases) and 15 new models across the LLM providers we track. The biggest cut was gemma-4-31b-it cached input, down 50% ($0.040 → $0.020 per 1M tokens).

The median cut this week was 20%. Figures are from our daily re-check of official provider pricing; see the live tracker for the full history.

12
Price changes
5
Price drops
7
Price increases
15
New models

Price cuts in Week of 31 August – 6 September 2026

Models that got cheaper, biggest cut first. Click a column to sort.

gemma-4-31b-itCached input$0.040/1M$0.020/1M-50%31 August 2026Google
Llama-3.3-70B-InstructInput$0.135/1M$0.100/1M-26%3 September 2026Meta
Llama 3.1 8b InstructInput$0.025/1M$0.020/1M-20%5 September 2026Meta
Mistral Medium 3.5 128BInput$1.65/1M$1.50/1M-9%3 September 2026Mistral AI
gemma-4-31b-itOutput$0.330/1M$0.300/1M-9%31 August 2026Google

Price increases in Week of 31 August – 6 September 2026

Llama 3.1 8b InstructOutput$0.025/1M$0.040/1M+60%5 September 2026Meta
Mistral Medium 3.5 128BOutput$5.50/1M$7.50/1M+36%3 September 2026Mistral AI
gemma-4-31b-itInput$0.080/1M$0.100/1M+25%31 August 2026Google
Claude Fable 5.1 (Bedrock)Input$10.00/1M$11.00/1M+10%5 September 2026AWS Bedrock
Claude Fable 5.1 (Bedrock)Output$50.00/1M$55.00/1M+10%5 September 2026AWS Bedrock
Claude Fable 5.1 (Bedrock)Cached input$0.250/1M$0.275/1M+10%5 September 2026AWS Bedrock
Claude Fable 5.1 (Bedrock)Cache write$12.50/1M$13.75/1M+10%5 September 2026AWS Bedrock

Models launched in Week of 31 August – 6 September 2026

Cite this recap

Monthly recaps are generated from our open price-change dataset (CC BY 4.0). See the recap JSON, the full change dataset, or the RSS feed.