Leaderboard / Model
GPT-6 Astra
by OpenAIAssistantAccess through OpenAI's ChatGPT plans — check current tier availability
OpenAI's GPT-6 Astra release, benchmarked in the arena writing one-shot Blender Python from the prompt text.
How to use GPT-6 Astra with Blender
Either paste its generated bpy scripts into Blender's Scripting workspace, or point an MCP-capable client running the model at the blender-mcp server for live scene access.
Best for: Procedural and parametric modelling driven from a script or a pipeline, where you want the strongest available reasoning applied to a single-shot build.
Which AI built it better?
Two random Season 3 entrants answered the same prompt. Pick the better build and the names are revealed. Your vote counts toward the leaderboard.
Picking a matchup…
What GPT-6 Astra built
All 12 Season 3 prompts, one script each, exactly as it came out. Click a build to orbit it in 3D, or compare every AI on the same prompt.
Treasure chestProps Desk lampProps Low-poly knightCharacters Mushroom creatureCharacters Stone watchtowerArchitecture Modern houseArchitecture Pickup truckVehicles Sailing boatVehicles Potted plantNature Island sceneNature Spiral staircaseAbstract Chess setAbstract
Earlier seasons stay published in the Season 1 archive.
Arena record
Season 3Ranked
- 1
- Entrants
- 12
- Builds
- of 12 prompts
- 2238
- Times voted on
- GPT-6 AstraProvisional1579 ±641228–839–119
- Season 2 final: GPT-6 Astra finished 1= (technical draw) on 1674 ±65. Full Season 2 results
Every entrant starts each season at 1500 ±350 and stays unranked until enough votes are in. Nothing carries over between seasons, ratings move only on real votes, and every ballot is published. How the rating works
What is GPT-6 Astra?
GPT-6 Astra is OpenAI's GPT-6 release. Like the other general assistants in the arena, it has no first-party Blender integration: for 3D work it either writes Blender Python (bpy) that you run in the Scripting workspace, or drives a live scene through an MCP-capable client connected to the open-source blender-mcp server.
It is listed separately from the ChatGPT profile because the arena competes per-model, not per-brand. GPT-5.6 Sol has its own Season 1 record under the same publisher, and keeping them apart is what makes it possible to see whether a model generation actually moved 3D output quality.
Astra entered Season 2 at extra-high reasoning effort, writing a single bpy script per prompt executed once in headless Blender 5.1. It covered all twelve prompts, and every one produced geometry — no failures to publish.
Strengths and limitations
Strengths
- Strength:Completed the full twelve-prompt set on the first run, with no empty or errored outputs
- Strength:Produces long, self-contained procedural scripts — its benchmark scenes run from 18 to 174 separate objects
- Strength:Uses Blender's own material system rather than flat colour, with 7 to 16 materials per scene
- Strength:Adjustable reasoning effort, so time can be traded for quality per task
Limitations
- Limitation:Ratings reset every season: its Season 2 record is archived, and it starts Season 3 uncalibrated like every other entrant
- Limitation:Not a mesh generator: geometry comes from code, which makes organic and sculpted subjects the structural weak spot
- Limitation:No official Blender connector; integration is via community MCP tooling or pasted scripts
- Limitation:Extra-high reasoning effort costs meaningfully more time and tokens per generation than the lower settings