Mozilla Publicly Opposes Google's Proposal to Standardize Chrome's Prompt API

The Battle Over Browser AI Standards Begins

A standardization battle over built-in browser AI capabilities is heating up. In 2025, Google proposed the Prompt API, designed to provide a unified JavaScript interface for calling local large language models integrated into the Chrome browser, with plans to push it as a W3C standard. However, the proposal has drawn public opposition from Mozilla developers, who argue the proposal has fundamental flaws in interoperability and model neutrality.

Prompt API: Google's Browser AI Ambitions

The core idea behind the Prompt API is to allow web developers to directly call browsers' built-in local AI large language models through a standardized JavaScript interface. In Chrome's desktop version, this built-in model is Google's own Gemini Nano. Users must first download the model before using the feature, and local devices must meet relatively high hardware requirements: at least 4GB of VRAM, 16GB of RAM, and no less than 22GB of available storage space on the drive where the browser is installed.

Google's intention is clear — by pushing the Prompt API toward W3C standardization, it aims to make it a web specification followed by all browser vendors, thereby making "built-in browser AI" one of the foundational capabilities of the web platform.

Mozilla's Three Core Objections

The Interoperability Dilemma

Mozilla developers pointed out in their statement that the biggest problem facing the Prompt API is "interoperability." Different large language models each have unique architectural characteristics, training data preferences, and behavioral patterns, meaning system prompts must be finely tuned for specific models.

The problem is that prompts optimized for one model may produce drastically different results on another, or even constitute "over-correction." This is entirely different from traditional Web APIs — traditional APIs have deterministic inputs and outputs, whereas LLM responses are inherently non-deterministic and highly dependent on the model's own characteristics.

If the Prompt API becomes a standard, prompts written by web developers will inevitably be optimized around Chrome's built-in Gemini Nano as the benchmark. When these applications run on Firefox or Safari, the user experience may differ significantly or even malfunction due to differences in the underlying model.

A Lose-Lose Choice for Competitors

Mozilla further pointed out that to achieve interoperability with Chrome, Mozilla and Apple would face an awkward choice: either obtain licensing for Google's model and integrate Gemini Nano or a compatible version into their own browsers, or independently develop and release a model that is highly compatible with Gemini Nano in behavioral characteristics.

Either way, competitors would be forced to build around Google's model ecosystem, effectively granting Google the power to define standards in the browser AI space.

The Absence of Model Neutrality

Another key issue raised by Mozilla is the lack of model neutrality. A true web standard should be vendor-neutral and should not favor any specific implementation. However, the Prompt API's design is inherently deeply tied to specific model behavioral characteristics, turning the "standard" into an endorsement of a particular vendor's technical approach. Unlike rendering engines or network protocols, which can ensure consistency through clear specifications, the ambiguity of LLM outputs makes truly neutral standardization nearly impossible.

A Deeper Industry Power Play

The essence of this dispute goes far beyond the technical level. At a time when AI has become a core competitive advantage in the tech industry, the standardization of AI capabilities in browsers — one of the most important application platforms — will profoundly shape the direction of the entire web ecosystem.

With Chrome commanding over 60% of global market share, Google wields enormous influence in driving web standards. Historically, Google has repeatedly leveraged Chrome's market dominance to push its own technical solutions into de facto standards. This Prompt API proposal is seen by critics as yet another attempt by Google to extend its AI advantages into the web standards arena.

For Mozilla, this is not just about choosing a technical path — it's about the already fragile competitive landscape of the browser market. If Google successfully pushes an API deeply tied to its own model as a standard, competing browsers like Firefox will be perpetually playing catch-up in AI experiences.

Looking Ahead

Currently, the Prompt API is still in the proposal stage and has a long way to go before becoming an official W3C standard. Mozilla's objections may prompt the W3C community to evaluate the proposal more carefully.

One possible compromise direction is designing a higher-level abstract API that encapsulates model differences at the browser implementation layer rather than exposing them to web developers. But this raises another question: if the level of abstraction is too high, would the API's practical value be significantly diminished?

The road to browser AI standardization is destined to be bumpy. Against a backdrop of intertwined interests, how to advance technology while maintaining the core principles of the open web is a challenge the entire industry must face together. This standards battle between Mozilla and Google may be just the opening act of a much larger contest in the era of browser AI.