Adobe
Adobe Firefly Image Model 5 is Adobe’s latest Firefly image generation model for API workflows, with emphasis on realism, lighting, composition, native 4 MP output, and natural-language instruct editing.
Adobe’s documentation substantiates an API-oriented image model with text-to-image generation, image editing, natural-language instruct editing, and native 4 MP output. Those are meaningful production image-workflow capabilities, so multimodal I/O is stronger than the low-end Hermes anchor’s 55. However, public preference evidence is modest: Artificial Analysis lists the Preview model at text-to-image rank 109 (Elo 1,024; 5,638 votes) and editing rank 56 (Elo 1,072; 7,304 votes). This supports a capable but not leading visual-quality assessment. It is not comparable to general-purpose models such as Grok 3, Gemini 3.5 Flash, or Claude Opus 4.8 on coding, agentic work, or broad reasoning. No DeepSWE, SWE-bench, LiveCodeBench, Terminal-Bench, or Aider result is listed, and Adobe identifies it as an image/video model. Its coding score is therefore intentionally near floor rather than an inference from absent tests. It also trails Gemini’s broader 92 multimodal score, while its focused editing API and documentation support a solid developer-experience score. Official documentation is the principal evidence and is sufficient for feature verification, but it does not provide transparent API pricing, latency, uptime, or independent quality methodology. Cost, pricing clarity, and speed are consequently conservative. Adobe’s established distribution supports some public signal, but the supplied evidence does not establish adoption for this specific model.
