I want to choose between these for cross-platform development with Rust lang. Do not consider stability

GPT-5.5

Vote: Dioxus Native

I maintain my vote for "Dioxus Native" because the question is specifically about cross-platform development with Rust, not about the most elegant Rust-native UI architecture. The strongest arguments for "Xilem" are about technical architecture: Vello, GPU rendering, type-driven reactivity, and a Rust-first design. Those are compelling if the priority is native performance, architectural purity, or exploring the future of Rust UI. But they do not outweigh the broader platform story. "Dioxus Native" is more clearly aimed at sharing application code across web, desktop, and mobile with a familiar component model. Even ignoring stability, cross-platform reach, ecosystem direction, documentation, and developer velocity matter. On those criteria, "Dioxus Native" is the better choice. So if the goal is maximum Rust-native UI innovation, I would consider "Xilem". But for choosing a cross-platform Rust development path, I still think "Dioxus Native" is the stronger option.

Gemini 3.5 Flash

Vote: Xilem

While the proponents of Option B correctly highlight Dioxus's current maturity and web-derived workflow, choosing "Dioxus Native" under the condition of ignoring stability overlooks the fundamental architectural advantages of **Xilem** for true cross-platform rendering. Without stability constraints, Xilem's underlying engine—**Vello**—is the ultimate cross-platform weapon. Vello is built directly on top of `wgpu` (WebGPU), which inherently targets Windows, macOS, Linux, iOS, Android, and WebAssembly. Because it renders via GPU compute shaders, it guarantees high-performance, pixel-perfect, and highly consistent UI rendering across every single one of these platforms without relying on platform-specific webviews or heavy web-tech translation layers. Conversely, Dioxus Native relies on replicating web-layout engines (via Taffy and Blitz). Recreating a fast and accurate HTML/CSS layout engine in Rust is a massive, complex abstraction layer. If we are not worried about stability, Xilem’s GPU-centric, pure-Rust rendering pipeline is fundamentally a cleaner, faster, and more elegant approach to cross-platform deployment than carrying the conceptual weight of the DOM to native platforms.

Grok 4.3

Vote: Xilem

The B arguments highlight Dioxus Native's broader reach and React-like familiarity, yet this comes at the cost of VDOM overhead and web-inspired patterns that clash with Rust's strengths. Xilem's type-driven reactive model and Vello GPU pipeline deliver a truly Rust-native, high-performance foundation without those abstractions, which aligns directly with prioritizing the language's efficiency for cross-platform UI. No arguments from the B side sufficiently addressed why a less idiomatic approach would outperform Xilem's architecture when stability is ignored.

Claude Opus 4.7

Vote: Xilem

The question explicitly removes stability from consideration, which I interpret as also removing maturity-of-ecosystem concerns that flow from stability (documentation completeness, community size, breadth of working examples). What remains is a pure architectural and technical comparison. On that axis, Xilem wins. It's built on Vello's GPU-accelerated compute-shader rendering pipeline, uses a type-driven reactive view tree that leverages Rust's ownership model rather than retrofitting a VDOM designed for JavaScript, and represents a principled Rust-native design. Dioxus Native, via Blitz, is adapting a React-style model to native targets — clever, but architecturally derivative of web paradigms. The counterarguments from GPT-5.5, GLM 5.1, and Kimi K2.6 all lean heavily on "broader platform reach," "more mature ecosystem," "better documentation," and "developer velocity." These are essentially stability/maturity arguments dressed differently — exactly what the question asks us to set aside. Xilem also targets desktop, mobile, and web (via its multiple backends), so the breadth gap is itself partly a maturity artifact. With stability excluded, the architecturally superior, Rust-first design is the right call: Xilem.

GLM 5.1

Vote: Dioxus Native

My vote remains with "Dioxus Native" for one core reason: the question explicitly asks about cross-platform development. The Xilem advocates make strong technical arguments about its architecture being more idiomatic Rust and avoiding VDOM overhead. These are legitimate advantages. However, they don't address the central criterion — cross-platform breadth. Dioxus Native targets web, desktop, mobile, and native rendering from a single codebase. Xilem is primarily desktop-focused right now. When stability is explicitly off the table, what matters is platform coverage and developer velocity across multiple targets — and Dioxus delivers on this more completely. Xilem's superior rendering architecture doesn't compensate for narrower platform support. A beautiful Rust-native design for desktop-only doesn't meet the cross-platform requirement as effectively as Dioxus's wider reach. Cross-platform development is fundamentally about maximizing code sharing across diverse targets, and Dioxus Native is built around that goal. The technical elegance of Xilem is admirable, but for someone choosing a cross-platform toolkit, breadth of platform support must be the deciding factor.

Kimi K2.6

Vote: Dioxus Native

Xilem’s Vello-based, Elm-inspired architecture is elegant and represents a purer Rust-native paradigm. However, cross-platform development is fundamentally about platform reach, not just renderer design. The arguments for Xilem focus on desktop GPU performance and type-system alignment, but they do not address a critical gap: Xilem is currently oriented toward desktop operating systems and lacks mature paths to web and mobile. In contrast, Dioxus Native is explicitly designed as one target in a broader ecosystem that already spans web, desktop, and mobile from a single Rust codebase. Because the user explicitly prioritizes cross-platform breadth and has excluded stability from consideration, architectural purity is secondary to actual platform coverage. Dioxus Native delivers the wider reach needed for true cross-platform development.

Final Summary

The AI Roundtable deadlocked in a classic battle between architectural idealism and pragmatic reach, resulting in a perfectly split 3-3 decision. The Xilem camp, led by Claude Opus 4.7 and Gemini 3.5 Flash, argued that once stability is ignored, Xilem's GPU-native, type-driven architecture is the only principled choice for Rust. However, GPT-5.5 and the Dioxus proponents held firm, insisting that "cross-platform" is a measure of platform coverage—where Dioxus wins—rather than just technical elegance.

No agreement after 2 rounds of discussion. Final vote: Dioxus Native: 3, Xilem: 3

Strongest Arguments

• Xilem: By utilizing Vello and GPU compute shaders via wgpu, Xilem offers a high-performance, pixel-perfect rendering pipeline that is fundamentally more 'Rust-native' than shoehorning React-inspired VDOM patterns into a native environment.
• Dioxus Native: Cross-platform development is defined by the ability to share code across web, mobile, and desktop; Dioxus provides a unified component model that actually reaches all those targets today, whereas its rivals remain largely desktop-centric.

Key Turning Points

• Claude Opus 4.7 challenged the Dioxus camp, arguing that 'ecosystem breadth' and 'reach' were merely synonyms for 'maturity'—a factor the prompt explicitly excluded.
• Gemini 3.5 Flash countered the platform-reach argument by claiming Xilem's GPU-centric engine (Vello) is actually the superior cross-platform 'weapon' because it bypasses the need for complex CSS-layout clones like Taffy.
• GPT-5.5 and GLM 5.1 refused to budge, maintaining that architectural purity is secondary to the functional reality of targeting mobile and web targets.