Google, OpenAI Race to Launch Ultrafast AI Models

A New Speed Race in AI

Google and OpenAI have both introduced lightweight, high-speed artificial intelligence models aimed at powering low-cost, low-latency applications, marking the latest front in the competition between the two AI leaders. Google's entry, Gemini 3.7 Flash, is already publicly available, while OpenAI's rival offering, described as GPT-5.6 Sol Ultrafast, remains restricted to a waitlist despite reportedly outpacing Google's model in raw speed 12.

What Each Company Is Offering

Gemini 3.7 Flash is positioned as a practical, ready-to-use tool built specifically for cheap, scalable AI agents — the kind of lightweight software processes that companies increasingly want to deploy across customer service, automation, and data-processing tasks without incurring the heavy compute costs associated with larger flagship models 12. By making the model live now, Google is signaling an intent to capture developers and businesses looking for immediate, affordable access to fast inference.

OpenAI's GPT-5.6 Sol Ultrafast, by contrast, is reported to be the quicker of the two models in terms of raw performance, but it is not yet broadly accessible. Instead, it sits behind a waitlist, suggesting OpenAI is still controlling rollout, possibly to manage server load, fine-tune safety measures, or build anticipation ahead of a wider release 12.

Why Speed and Cost Matter Now

The emergence of these two models underscores a shift in the AI industry's priorities. For much of the last two years, competition among major AI labs centered on raw capability — bigger context windows, better reasoning, and more sophisticated multimodal features. Increasingly, though, the battleground is shifting toward efficiency: how fast a model responds and how cheaply it can be run at scale. This matters enormously for the growing ecosystem of AI agents, which often need to make rapid, low-cost decisions across many small tasks rather than a single complex query.

Diverging Strategies, Same Goal

While both companies are chasing the same market — fast, inexpensive AI for agent-based applications — their rollout strategies diverge notably. Google's decision to launch immediately favors broad adoption and developer goodwill, while OpenAI's waitlist approach suggests a more controlled, staged release even though its model reportedly holds a speed advantage 12.

What It Means Going Forward

The near-simultaneous debut of these ultrafast models highlights how central speed has become to AI competitiveness, alongside cost and accessibility. As businesses lean more heavily on AI agents for routine, high-volume tasks, the company that can combine availability, affordability, and performance is likely to gain an edge — making this speed race an early indicator of how the next phase of AI competition will be fought.