OpenAI makes Decisions API, its Jev competitor, generally available with new voice and visual modalities, at $0.10/1M input tokens and no output-token charges

The cost to make AI model inferences plummets, potentially making new AI applications feasible for the first time.

OpenAI has made its Decisions API, a competitor to TypeSafe's Jev model, generally available. This API is designed for fast, low-cost decision-making tasks, acting as a super classifier that can quickly select options and provide confidence ratings. Originally announced on September 29th, the API is now accessible to all customers. OpenAI reconfigured its ChatGPT Luna model to achieve significantly faster inference times, reducing latency from seconds to milliseconds, and set its pricing at $0.10 per 1 million input tokens with no output-token charges. The Decisions API also includes new voice and visual modalities and has been geographically optimized for lower latency. Early adopters like Box, Datadog, Ramp, and Speak have reported replacing their existing OpenAI LLMs with the Decisions API for higher throughput and reduced action times, with Speak noting a reduction from 1.4 seconds to 0.13 seconds.

OpenAI's move to generalize its Decisions API signals a heated market battle in the specialized domain of rapid, cost-effective AI decision-making. By leveraging and reconfiguring its existing Luna model, OpenAI aims to directly challenge startups like TypeSafe, which has carved out a significant niche with its Jev model. This competition is pushing the boundaries of speed and affordability, indicating a broader industry trend toward optimizing AI for specific, high-throughput tasks rather than relying solely on general-purpose LLMs for all functions. The introduction of new modalities and aggressive pricing suggests that OpenAI is not only seeking to capture market share but also to define the standard for this emerging category of AI.

The widespread availability and low cost of OpenAI's Decisions API, coupled with its new multimodal capabilities, is poised to democratize the creation of AI-powered decision-making tools. Developers can now integrate AI into applications where immediate, low-cost choices are critical, such as real-time analysis of visual data or instant voice command processing. This competitive pressure will likely drive further innovation and price reductions across the sector, potentially making sophisticated AI decision agents a standard component in a wider array of software and services, accelerating the adoption of AI in everyday technology.