Executive Summary
The July 2026 AI Gateway Production Index reveals a rapidly bifurcating enterprise AI market amidst strong overall growth (29% token volume increase MoM). Open-weight models, particularly DeepSeek, have surged to capture nearly a third of all token volume on less than 4% of spend. However, frontier labs like Anthropic continue to dominate enterprise budgets, securing 95% of total spend by focusing on high-stakes use cases like coding and back-office agents. This demonstrates an emerging enterprise strategy of cost-optimizing workloads by routing them to the most appropriate model class.
Key Takeaways
* Market Growth & Price Stability: Total token volume grew 29% month-over-month, with spend increasing 27%. The average price per token held flat, as the rise in cheap open-weight volume was offset by price increases for frontier models.
* Open-Weight Surge: Open-weight models now process 29% of all production tokens, nearly tripling their share since April, while accounting for just 4% of spend.
* Key Player Shifts: DeepSeek is now the third-largest provider by token volume at 22.6%, nearly catching Google (24%). Z.ai's new open-weight model, GLM 5.2, showed exceptionally rapid adoption after its mid-June release.
* Frontier Dominance in Spend: Anthropic alone captured 61% of market spend on 32% of tokens, maintaining its lead in high-value workloads. OpenAI's token share fell, but its spend share rose, indicating a shift toward more costly tasks.
* Fragmented Modality Leadership: No single lab leads across all modalities. Anthropic leads text, OpenAI leads images (53% of generations), xAI leads video volume (Grok Imagine, 42%), and Chinese labs lead video spend.
* Regulatory Impact: Anthropic's new Claude Fable 5 model saw rapid adoption upon release but was temporarily suspended for several weeks to comply with a new US export-control directive.
Strategic Importance
This report shows the enterprise AI market is maturing beyond using a single model for all tasks. Companies are actively implementing sophisticated routing strategies to optimize for cost and capability, creating distinct, parallel markets for both low-cost open-weight models and premium frontier models.