GPT-Image-2.5 (Sunburst) vs Union Alpha (Stealth)

A direct, empirical evaluation of GPT-Image-2.5 (Sunburst) (OpenAI) against Union Alpha (Stealth) (OpenRouter & OpenCode).

GPT-Image-2.5 (Sunburst)

Developed by OpenAI · Proprietary Multimodal DiT parameters

High-fidelity typographic rendering, raytracing reflection accuracy, sketch-to-image conditioning, and localized comment inpainting.

Union Alpha (Stealth)

Developed by OpenRouter & OpenCode · Multi-Model Ensemble Mixture parameters

Terminal coding agents, pull-request synthesis, multi-model consensus verification, and zero-cost autonomous refactoring.

Metric GPT-Image-2.5 (Sunburst) Union Alpha (Stealth)
SWE-bench Verified 62.4% 74.2%
MMLU-Pro 89.5% 89.8%
MATH-500 91.2% 95.4%
Context Window Spatial Vector Token Canvas (1792×1024) 256,000 tokens
Input Pricing (per 1M) $2.50 $0.00
Output Pricing (per 1M) $10.00 $0.00

Architectural & Practical Deployment Analysis

Evaluating GPT-Image-2.5 (Sunburst) (OpenAI) alongside Union Alpha (Stealth) (OpenRouter & OpenCode) highlights key technical trade-offs across neural parameter scale and compute efficiency. GPT-Image-2.5 (Sunburst) executes across Proprietary Multimodal DiT total parameters with Continuous Flow Transformer active per token, whereas Union Alpha (Stealth) operates on Multi-Model Ensemble Mixture parameters with Dynamic MoA Router active per token.

On SWE-bench Verified software engineering tasks, GPT-Image-2.5 (Sunburst) achieves 62.4% autonomous resolution rate, compared to 74.2% for Union Alpha (Stealth). On complex reasoning benchmarks, GPT-Image-2.5 (Sunburst) posts 89.5% on MMLU-Pro and 91.2% on MATH-500, against 89.8% and 95.4% respectively for Union Alpha (Stealth).

From an operational economics perspective, processing 10,000,000 mixed tokens (7M input / 3M output) costs approximately $47.50 on GPT-Image-2.5 (Sunburst) versus $0.00 on Union Alpha (Stealth).

Recommendation: Deploy GPT-Image-2.5 (Sunburst) when your infrastructure prioritizes high-fidelity typographic rendering, raytracing reflection accuracy, sketch-to-image conditioning, and localized comment inpainting.. Deploy Union Alpha (Stealth) when workload requirements favor terminal coding agents, pull-request synthesis, multi-model consensus verification, and zero-cost autonomous refactoring..