Dogfight
Dogfight — air-combat shooter.
What I asked each model — the Dogfight prompt
Every model on this page got this exact prompt inside the Agent Operating System: Dogfight — air-combat shooter.
Single HTML file out. No iteration. No examples in the system prompt. Whatever each model produced on the first run is what's on this page. 24 frontier models have attempted it so far: Claude Fable 5, Fugu Ultra, Fugu Ultra 1.1, Fugu Mini, Fusion, Gemini 3.6 Flash, GLM-5.2, GPT-5.6 Sol, Grok, Inkling, Kimi K3, MiniMax M3, Hermes MoA, Opus 4.8, Claude Opus 5, Qwen 3.8, Qwen 3.7, Claude Sonnet 5, DeepSeek V4 Pro, DeepSeek V4 Flash, Kimi K2.7, Kimi K2.7 · Fast, Kimi K2.7 · No-Think, Kimi K2.7 · Quality.
Why this task matters. Dogfight is a textbook test of game-class capability — the kind of build that exposes whether a model is doing pattern-matching or actual reasoning. A model that ships this in one shot is usually safe to wire into your agent loop for harder tasks of the same shape.
How each model handled Dogfight
Ranked by my 0–10 score from the source comparison guides on agentos.guide. Click any to play the actual one-shot HTML the model produced.
What I saw: Renders cleanly with a solid 3D scene—chase-cam plane, clouds, mountains, HUD, health bar and crosshair all present and on-brief. But no enemies visible in the shot and the plane looks slightly low/generic; competent and shippable yet not distinctive enough to top the field.
What I saw: Ultra v2 (gap-fill) — 3D dogfight. Smoke-test MAYBE (0.0% diff) — flight controls may need specific keys; flagged for manual verification.
What I saw: Strong visuals — detailed hero jet with shadow, polished HUD, city skyline and terrain create a real air-combat atmosphere; but the screenshot shows no visible enemies or combat (0/10 kills, no dogfight action), so it reads closer to a flight-sim showcase than a proven shooter.
What I saw: Dogfight scene structurally complete but smoke-test shows MAYBE-STATIC (<0.5% pixel diff after input) — possible the plane isn't taking keys reliably.
What I saw: RETRY @ 24K tokens — now complete: 36KB three.js + WebGL with 2 rAFs, 6 input handlers, full </html>. Fighter, enemies, missiles, guns, sky + ground, full HUD. The original truncated 32KB attempt has been replaced with a working build.
What I saw: Gorgeous low-poly 3D scene with a detailed player jet, glowing engines, volumetric clouds, terrain, and a full sci-fi HUD (hull, boost, radar with hostile blips, target-lock reticle). Enemies are present on radar and in-world but combat isn't clearly shown mid-fight in the shot, keeping it just shy of the very top.
What I saw: Beautifully rendered third-person chase-cam scene with a detailed player jet, layered clouds, moon, enemy squadron inbound, and a cohesive HUD (crosshair, armor/boost meters, working radar with blips). Cinematic art direction and complete combat framing make it a clear task winner; only minor risk is depth in actual combat feel.
What I saw: 3D dogfight game with fighter jet, enemies, missiles + guns, sky + ground, crosshair + speed + altitude HUD. 30KB.
What I saw: Clean neon HUD, functional core loop (mouse-follow, shooting, spawns, collisions/lives), but visually thin — the player 'plane' reads as a flat pink block and clouds are drab boxes with no dogfight drama or enemies visible in frame. Works and is polished at the UI level but generic and low-fidelity for the 3D combat feel.
The winner on Dogfight
Fusion took gold on this task. winner · air combat.
What I saw: RETRY @ 24K tokens — now complete: 36KB three.js + WebGL with 2 rAFs, 6 input handlers, full </html>. Fighter, enemies, missiles, guns, sky + ground, full HUD. The original truncated 32KB attempt has been replaced with a working build.
See Fusion's full model card: /models/fusion. Direct head-to-head against the runner-up: Fusion vs GPT-5.6 Sol.
Every attempt — live, playable
Side by side. Click any tile to run that model's actual one-shot HTML in a new tab.
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVE
▶ LIVEHow I scored Dogfight — methodology
Three axes, 0–10 each, averaged. Runs: drop the .html in a browser; if it opens to a broken page, it scores zero. Hits the brief: did the model ship the thing the prompt asked for, or a different thing it found easier. Looks good: visual polish, motion, interactivity — where most of the gap between gold and silver lives.
My scores trace back to the source comparison guides on agentos.guide. See the full methodology page for data provenance, including which source guide each cell's score came from.
Related
More game benchmarks: all tasks in the Game category · See the best AI model for Dogfight · Back to the leaderboard
Run this stack yourself.
Every demo on this bench was built inside the Agent Operating System — one prompt, one shot, single HTML file out. The Agent OS, the prompts, the templates, the weekly walkthroughs and 4,000+ founders shipping with it every day all live inside the AI Profit Boardroom.