AI News·8 Nov 2026·via Dev Community

    Frontier LLM prices sat flat all October. Then two cheaper flagships took seats, and one doubles on December 1

    The LLM Price Index, which tracks the blended price of ten flagship models, held steady at $5.14 per million tokens throughout October, marking its longest recorded stable period. However, on November 7, the index fell 4.9% to $4.89 per million tokens, following the official inclusion of Google's Gemini 4 Argon and Moonshot's Kimi K3.1 Pro, both launched at lower prices than their predecessors.

    The recent stability in the LLM Price Index, an equal-weight average of ten flagship models' blended prices, concealed market shifts. Despite the index remaining flat through October, Google's Gemini 4 Argon and Moonshot's Kimi K3.1 Pro were introduced at a blended $4.00 per million tokens. These new models formally replaced their higher-priced predecessors in the index on November 7, resulting in the reported 4.9% price reduction.

    This apparent cost reduction is largely temporary. Gemini 4 Argon's $4.00 blended price is an introductory offer valid only until November 30. From December 1, its list price will increase to $8.00 per million tokens, a 78% rise over the model it replaced. This single repricing event is projected to add $0.40 to the index, raising it to $5.29, which would reverse November's decline and place the index 2.9% above its October level. This highlights how time-limited promotions can distort overall market price perception.

    Beyond index fluctuations, a broader trend shows LLM providers primarily reduce costs by launching new models rather than repricing older ones. For instance, Claude Haiku 5.5 offers a 90% cost reduction for prompts up to 100K tokens compared to its predecessor, Haiku 4.5. Simultaneously, several labs, including OpenAI, Alibaba, Google, and Mistral, retired or deprecated numerous models in October. OpenAI's suggested replacements for its smaller retired models could lead to cost increases of 2.3x to 6x, underscoring the need for users to assess replacement costs and explore alternatives like newer, cheaper models in a line.

    Share this article

    Go deeper
    Original reporting by Dev CommunityWe don't republish, read the full story →

    Related reading

    6 stories