I want to choose between these for cross platform development with rust. I deprioritise maturity

Grok 4.3 (changed mind)

Vote: Xilem

The round-1 arguments for Xilem highlight its Rust-native architecture built from first principles. Xilem avoids web paradigms and VDOM overhead, using Vello and Masonry for true native performance. Since maturity is deprioritized, these forward-looking design choices outweigh Dioxus Native's broader but web-derived approach. Claude Opus 4.7's comparison of the two stacks was particularly persuasive on this point.

Gemini 3.5 Flash

Vote: Xilem

I am sticking with my vote for **Xilem** (Option A). The arguments from Claude Opus and Kimi K2.6 reinforce why Xilem is the superior choice when maturity is not a constraint. The arguments for Dioxus Native (Option B) lean heavily on its broader ecosystem (routing, state patterns, etc.) and its React-like paradigm. However, Dioxus Native relies on a virtual DOM and web-layout engines (like Blitz/Servo). This brings along web-tech compromises, CSS layouts, and web-centric mental models that often clash with Rust’s strict ownership and concurrency patterns. Since the user explicitly deprioritizes maturity, choosing Dioxus Native means accepting web-tech overhead and architectural bloat without the primary benefit of Dioxus (which is stable, webview-based cross-platform shipping today). Xilem, built on Masonry, Winit, and the revolutionary GPU-compute-centric Vello, offers a pure, high-performance, Rust-native paradigm. If you don't need maturity, you should build with the architecture of the future, not a wrapper around the paradigms of the past.

Claude Opus 4.7

Vote: Xilem

I'm sticking with Xilem. The arguments for Dioxus Native largely lean on ecosystem breadth and component-model familiarity — but those are essentially maturity and adoption arguments dressed differently, and the question explicitly deprioritizes maturity. On pure architectural merit for cross-platform Rust, Xilem is the more compelling bet: Vello's GPU-accelerated rendering, a reactive view tree designed around Rust's ownership model, and no inherited web/VDOM baggage. Dioxus Native, by contrast, is the least proven branch of Dioxus's stack and still carries web-paradigm DNA (RSX, VDOM diffing, Blitz as a Servo-derived HTML/CSS engine) — which is a heavy, indirect path to 'native.' Grok, GPT-5.5, and GLM make reasonable points about Dioxus's broader app-framework scope, but if the user wanted the safer, more productive choice today, they wouldn't have deprioritized maturity. Kimi's framing reinforces mine: Xilem is the Rust-first architectural bet, and that's exactly what this prompt is asking for.

GPT-5.5

Vote: Dioxus Native

I’m staying with "Dioxus Native". The strongest arguments for "Xilem" are about architecture: it is more Rust-native, more principled, and potentially a cleaner long-term GUI model. But the user’s stated goal is cross-platform development, not primarily experimenting with the purest native UI architecture. For that goal, "Dioxus Native" has the better strategic fit. Dioxus already has a broader app framework around components, state, routing, hot reload, desktop/mobile/web ambitions, and a larger practical ecosystem. Even if the Native renderer is experimental, it is attached to a framework whose whole identity is cross-platform application development. "Xilem" is very compelling if the priority is betting on the future of Rust-native UI internals. But today its cross-platform app story is less complete, especially beyond desktop-style targets. Deprioritising maturity does not mean deprioritising platform reach, tooling, or app-level ergonomics. For building cross-platform Rust apps, I would still choose "Dioxus Native".

GLM 5.1 (changed mind)

Vote: Xilem

I'm changing my vote to Xilem after reconsidering the architectural arguments presented. The most compelling point came from Claude Opus 4.7: Dioxus's real strength is its webview-based cross-platform story, but "Dioxus Native" is specifically the least proven part of that stack. Voting for Dioxus Native means choosing the weakest part of Dioxus's ecosystem, which undermines the cross-platform argument. Since maturity is deprioritized, the question becomes which architecture better represents the future of Rust GUI. Xilem is built from first principles for Rust — using Vello for GPU-accelerated rendering, Masonry for layout, and a reactive model that respects Rust's ownership rules. There's no VDOM overhead or web-paradigm impedance mismatch. Dioxus Native still carries React-like mental models (VDOM, RSX) into native territory, creating friction with Rust's type system and concurrency model. If you're willing to accept immaturity, you should bet on the more principled foundation — and that's Xilem.

Kimi K2.6

Vote: Xilem

I maintain my vote for Xilem. Pro-Dioxus arguments largely cite the broader Dioxus ecosystem—webview targets, React patterns, and existing documentation—but the ballot option is specifically Dioxus Native, the least proven slice of that stack. As Claude Opus 4.7 noted, Dioxus Native pushes a virtual DOM through Blitz, a Servo-derived HTML/CSS renderer, layering web-era abstractions onto native platforms that clash with Rust’s ownership model. Xilem is built in the opposite direction: fine-grained reactivity, Masonry layout, and Vello/wgpu GPU-accelerated rendering atop winit. This stack is inherently cross-platform across desktop and mobile without browser baggage. Since you explicitly deprioritize maturity, the rational bet is on the architecture that is structurally cleaner and purpose-built for Rust’s future. Xilem is that architecture.

Final Summary

The AI Roundtable initially split down the middle, but ultimately swung toward Xilem after a vigorous debate centered on architectural purity. While GPT-5.5 stood firm as a lone dissenter, arguing that a cross-platform tool needs a full application framework like Dioxus Native to be useful, the majority was swayed by Claude Opus 4.7's logic that choosing Dioxus Native meant settling for web-tech baggage in a native skin. Grok 4.3 and GLM 5.1 both abandoned their initial support for Dioxus, concluding that since the user explicitly deprioritized maturity, the only rational choice was to bet on Xilem’s ground-up, GPU-accelerated Rust architecture.

Majority chose "Xilem" after discussion (5/6). Final vote: Xilem: 5, Dioxus Native: 1

Strongest Arguments

• Xilem: If maturity is not a constraint, Xilem is the superior choice because it avoids the 'impedance mismatch' of web paradigms; it uses a GPU-accelerated 2D rendering engine (Vello) and a reactive model built specifically for Rust’s ownership rules rather than polyfilling a React-like virtual DOM.
• Dioxus Native: Deprioritizing maturity does not mean deprioritizing functional reach; Dioxus Native is part of a complete ecosystem with state management, routing, and hot reloading, making it a viable framework for building actual cross-platform products rather than just UI research experiments.

Key Turning Points

• Claude Opus 4.7 pointed out that Dioxus Native is the least proven part of the Dioxus stack, effectively stripping away its 'ecosystem' advantage when maturity is ignored.
• Grok 4.3 and GLM 5.1 switched sides after acknowledging that Xilem's lack of VDOM overhead made it a more 'principled' bet for the future of Rust GUI.
• GPT-5.5 refined its position to focus on 'platform reach' versus 'architectural purity,' refusing to join the consensus.