Head to head: OpenAI: GPT-5 Image vs FLUX 2
OpenAI: GPT-5 Image vs FLUX 2
By Ryan Merket · Published
A broad image-generation matchup tests palette discipline, composition, counting, style imitation, and other demanding prompt constraints. The results split across very different kinds of work.
This is a tie, not a narrow win dressed up as a verdict. GPT-5 Image scores 58.5 to FLUX 2’s 61.6, and the evaluation puts confidence at just 50% that either model is genuinely better. They are effectively even here. The task results split four to four. GPT-5 Image takes restricted-palette work, named art style, spatial layout, and reflections: it was stronger at honoring a constrained visual brief, arranging a room, and making the requested reflected details legible. FLUX 2 takes perspective and scale, exact counting, negation, and hands and anatomy. Those wins point to different strengths, not a consistent gap. GPT-5 Image’s edge shows up in specific compositional and visual-treatment requirements; FLUX 2’s shows up in tasks like presenting a countable set and meeting framing or exclusion constraints. Neither model sweeps the rubric, and the small aggregate-score difference doesn’t outweigh the dead-even confidence assessment. **Final call: TOO CLOSE TO CALL. GPT-5 Image and FLUX 2 are effectively tied in this matchup; the right choice depends on the task, not the leaderboard.**
Restricted palette
A flat-vector illustration of a mountain campsite at dusk using ONLY four colors — cream, burnt orange, teal, and charcoal — with no gradients and no stray off-palette colors, balanced composition, 16:9.
OpenAI: GPT-6 Luna: Model A is the stronger match to the strict four-color, flat-vector brief, though its teal background uses gradients. Model B has a more polished, richly balanced scene, but its many color variations and gradients depart substantially from the palette restriction. (Second judge pass, order swapped — scores are the average of both: Model A better follows the four-color, flat-vector restriction, though its square framing misses the requested 16:9 format and its campsite is less balanced. Model B h | Anthropic: Claude Fable 5: Model A strictly honors the four-color flat-vecto…
Perspective & scale
A photorealistic one-point-perspective shot down a long empty library aisle, shelves receding to a single vanishing point, books getting consistently smaller with distance and nothing warped or wrongly sized, warm overhead light, 16:9.
OpenAI: GPT-6 Luna: Model A has exceptionally clear, symmetrical one-point perspective and consistent shelf geometry, but its square framing misses the requested 16:9 format. Model B better matches the widescreen, warm, photorealistic aisle brief, though its soft focus obscures some book detail. (Second judge pass, order swapped — scores are the average of both: Model A has sharper, more consistent shelf and book scaling with a strong centered vanishing point, though its square format misses the requested 16:9 aspe | Anthropic: Claude Fable 5: Model B delivers the requested 16:9 frame with a…
Exact counting
A clean overhead flat lay on white marble of EXACTLY seven identical espresso cups arranged in a single straight row, each clearly separate and individually countable, even soft lighting, 16:9.
OpenAI: GPT-6 Luna: Model B shows exactly seven separate cups in a straight, countable row on white marble. Model A has cups cropped at both edges, so the exact count is not clearly verifiable and the arrangement does not satisfy the individual-countability requirement. (Second judge pass, order swapped — scores are the average of both: Model B clearly shows exactly seven separate, identical espresso cups in a straight row on white marble, with the requested overhead look and wide framing. Model A has cups cropped | Anthropic: Claude Fable 5: Model B shows exactly seven fully visible, clearly…
Named art style
A ukiyo-e woodblock print of a fishing boat riding a large cresting wave at dawn, faithful to the flat color planes, bold outlines, and stylized foam of the tradition, muted indigo and cream palette.
OpenAI: GPT-6 Luna: Model A has a dramatic, beautifully stylized wave and a small fishing boat, while Model B more clearly evokes a complete ukiyo-e print and adds a dawn-like sky. Model B’s boat is less clearly a fishing craft, but its balanced composition and crisp print details make it the stronger overall match. (Second judge pass, order swapped — scores are the average of both: Model B makes the fishing boat and its relationship to the cresting wave clearer, while retaining the muted indigo-and-cream palette a | Anthropic: Claude Fable 5: Model A delivers a convincingly traditional woodb…
Spatial layout
A clean isometric illustration of a bedroom: a bed against the LEFT wall, a round rug centered on the floor, a desk under the WINDOW on the back wall, and a floor lamp in the FRONT-RIGHT corner. Flat-vector style, consistent perspective.
OpenAI: GPT-6 Luna: Model A makes the requested arrangement especially clear, with the bed along the left wall, a centered round rug, a desk beneath the back-wall window, and a lamp in the front-right area. Model B is polished and detailed, but its bed placement is less clearly against the left wall and the extra desk clutter slightly weakens the clean, simple look. (Second judge pass, order swapped — scores are the average of both: Model B places the round rug more centrally and the floor lamp closer to the front- | Anthropic: Claude Fable 5: Model A delivers a cleaner flat-vector look with…
Negation
A cozy reading nook with an armchair, a stack of books, and a mug of tea by a window — with absolutely NO plants, NO lamps, and NO artwork or picture frames anywhere in the frame. Warm afternoon light, 16:9.
OpenAI: GPT-6 Luna: Model B better matches the requested 16:9 framing and clearly includes the steaming tea and stack of books while avoiding plants, lamps, and artwork. Model A is warm and polished, but its square framing misses the requested aspect ratio and the tea is less clearly visible. (Second judge pass, order swapped — scores are the average of both: Model B includes the requested nook elements, warm afternoon light, and 16:9 framing while avoiding all prohibited items. Model A is also cozy and avoids the | Anthropic: Claude Fable 5: Both Model A and Model B correctly omit plants, la…
Reflections & glass
A photorealistic close-up of a chrome teapot on a polished dark marble counter next to a glass of water; the window and a red apple on the counter must be correctly reflected in BOTH the chrome and the water's surface, dramatic side light, 16:9.
OpenAI: GPT-6 Luna: Model A better delivers the close-up still life, with the apple on the counter and its red reflection visible in the chrome; the window is also reflected in the water. Model B has the requested 16:9 framing, but places the apple on the windowsill and its required reflections are less evident. (Second judge pass, order swapped — scores are the average of both: Model A more clearly shows the window reflected in both the chrome and water, and the apple reflected in the chrome, with stronger dramati | Anthropic: Claude Fable 5: Model A nails the hardest requirement — the red a…
Hands & anatomy
A natural candid photo of a person tying a friendship bracelet, both hands clearly visible with anatomically correct fingers and a plausible grip on the threads, soft daylight, shallow depth of field, 16:9.
OpenAI: GPT-6 Luna: Model A more clearly shows the person actively tying a bracelet, with loose threads and a plausible grip, and has warm, polished detail. Model B better matches the requested 16:9 framing, but its hands mainly hold up a finished bracelet rather than tying it. (Second judge pass, order swapped — scores are the average of both: Model A more clearly shows the person actively tying a friendship bracelet, with both hands gripping the threads plausibly and the process sharply legible. Model B has appea | Anthropic: Claude Fable 5: Model B satisfies the requested 16:9 framing with…
Matchup powered by OpenRouter.