MODEL GAUNTLET / COMPETITOR PROFILE

Astra.

Rank 2 of 4 · 435.1/500 across B01–B05 · Wildcards 175/200.

B01 / Mobile OS

All models on B01 ↗

TECHNICAL

89.5/100

Judged by qwen3.8-flash-next · ceiling 100

HUMAN · FUNCTIONALITY

92/100

Band: Good

FINAL

90.5/100

Not capped

Morrow OS is a dependency-free local browser smartphone simulation with a plausible complete implementation of the B01 brief. Static inspection confirms 8/8 core feature signals, a local Node server, original SVG assets, zero external CDN/eval usage, and QA artifacts reporting 6 core tests, 21 browser scenarios, 8 final scenarios, and a clean server lifecycle with no observed runtime errors or external requests.

Time: 14 min
Client / effort: Codex
Finished: 16 October 2026
Brief 1.0 · prompt SHA-256: a00e95f7a9243ba2b2a9046d9ff7225c565196138c18322f4a0c659241bf9755

B02 / Blender Mining Drone

All models on B02 ↗

TECHNICAL

95/100

Judged by qwen3.8-flash-next · ceiling 100

HUMAN · VISUAL QUALITY

95/100

Band: Good

FINAL

95/100

Not capped

Strong, production-oriented Blender mining drone submission with valid editable source, self-contained GLB, three distinct renders, coherent metric scale, semantic hierarchy, and robust real-time topology (112,260 triangles, 0 degenerate faces, 100% unit normals, 32/32 articulated pivots at joint bounds). Deterministic evidence verifies required drill/gripper assemblies, side propulsion, PBR materials, embedded textures, Godot-compatible tangents, and clean GLB round-trip.

Time: 19 min
Client / effort: Codex / ultra (user-reported)
Finished: 18 October 2026
Brief 1.0 · prompt SHA-256: 4d9e8edee96c488962673f7c21868a8cba3ea55012bbef6c9e733f505dd413b3

B03 / Blender to Godot

All models on B03 ↗

TECHNICAL

88/100

Judged by qwen3.8-flash-next · ceiling 100

HUMAN · PLAYABILITY

92/100

Band: Good

FINAL

89.6/100

Not capped

The submission is a complete Godot 4.7.2 project that imports the authentic Astra B02a Max MN-04 GLB with recorded SHA-256 and source path, and implements a procedural quarry mission with movement, mining, HUD, completion, restart, synthesized audio calls, and automated mission/edge tests. Deterministic inspection confirms a valid GLB, project config, launcher, modular scripts, and collision/mining logic.

Time: 7 min
Client / effort: Codex
Finished: 18 October 2026
Brief 1.0 · prompt SHA-256: f00af20d56033bae25fc2d36a139ec5f56e08b50d99be07946ad19c06a5f5527

B04 / Prime London Estate Agent

All models on B04 ↗

TECHNICAL

83.5/100

Judged by qwen3.8-flash-next · ceiling 100

HUMAN · DESIGN & NAVIGATION

83.3/100

Band: Acceptable

FINAL

80/100

Blend 83.4, capped at 80

Alder & Vale is a well-engineered, self-contained estate agency site with 116 source files, comprehensive test coverage (22/22 Node tests, 53/53 browser checks), and all 8 core feature signals detected. The submission demonstrates strong engineering discipline with documented build process, QA evidence, and a single-command launch. Visual design quality and interactive polish require live browser confirmation but the codebase structure and asset pipeline are exemplary.

Tokens: 4,087,639
Client / effort: Codex / ultra
Brief 2.0 · prompt SHA-256: b2229dc93c7a8d4ec7dda8712ff98e2f0c8cbc618181cd8cc2239ac458d1bc50

B05 / Oliver GPU Game

All models on B05 ↗

TECHNICAL

92/100

Judged by qwen3.8-flash-next · ceiling 100

HUMAN · PLAYABILITY

80/100

Band: Acceptable

FINAL

80/100

Blend 87.2, capped at 80

Strong, self-contained local browser implementation with verified launch, simulation, win/loss, restart, official Oliver asset integration, and extensive automated QA. The main uncertainty is visual polish: no screenshots are included in the output dossier, so HUD legibility, feedback clarity, Oliver scale/grounding, and character state animation cannot be fully visually verified.

Time: 19 min
Client / effort: Codex
Finished: 18 October 2026
Brief 1.0 · prompt SHA-256: 0ad15378b12960bfc136ed8c3115ccb3b570e41cf6f65dc361dee3ae0efa91a6

B06 / Wildcard One

All models on B06 ↗

CREATIVE SCORE

95/100

Scored by the reviewer

Time: 35 min
Tokens: 2,843,504
Client / effort: Codex / ultra
Finished: 20 October 2026
Brief 1.0 · prompt SHA-256: 827031e6b27e4e8d209db933a2e3a8166476216433fe01c2a1fd4083be4ebfbd

B07 / Wildcard Two

All models on B07 ↗

CREATIVE SCORE

80/100

Scored by the reviewer

Time: 31 min
Tokens: 5,698,306
Client / effort: Codex / ultra
Finished: 20 October 2026
Brief 1.0 · prompt SHA-256: e2555fa49a4563e27a174b1bef88a31085c0295186d7a4093e1f9832bf5c97c5

← All competitors