Leaderboard / Model
GPT-6 Astra
by OpenAI · assistant · Access 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.
Arena Status
Season 2 — Calibration
Competitors
1
Benchmark Outputs
12
of 12 prompts
Ballot Appearances
0
| Competitor | Status | Rating | W–L–T |
|---|---|---|---|
| GPT-6 Astra | Calibrating | 1500 ±350 | 0–0–0 |
Every competitor starts each season at the Glicko-2 default of 1500 ±350 and stays unranked until it clears the calibration floors — no rating carries over between seasons. Ratings move only on real head-to-head votes: no seeded or editorial vote data is ever shown as community results, and every ballot is published. Read the methodology.
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
- +Completed the full twelve-prompt set on the first run, with no empty or errored outputs
- +Produces long, self-contained procedural scripts — its benchmark scenes run from 18 to 174 separate objects
- +Uses Blender's own material system rather than flat colour, with 7 to 16 materials per scene
- +Adjustable reasoning effort, so time can be traded for quality per task
Limitations
- −No blind-vote record yet — it entered at the start of Season 2, so its rating is uncalibrated and nothing here is a quality claim
- −Not a mesh generator: geometry comes from code, which makes organic and sculpted subjects the structural weak spot
- −No official Blender connector; integration is via community MCP tooling or pasted scripts
- −Extra-high reasoning effort costs meaningfully more time and tokens per generation than the lower settings
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.
Season 2 Benchmark Outputs
GPT-6 Astra's unedited results on the benchmark prompt set. Each competitor generated its entries independently — no shared code or outputs. Earlier seasons stay published in the archive.
P01: “A wooden treasure chest with iron fittings and a curved lid, slightly weathered, closed.”
P02: “A desk lamp with an articulated two-joint arm and a round metal shade, in a bent working pose.”
P03: “A low-poly stylized knight character in idle pose, with helmet, shield on back, T-pose acceptable.”
P04: “A cute cartoon mushroom creature with big eyes and stubby legs, standing.”
P05: “A small medieval stone watchtower with a wooden door, arrow slits, and a conical roof.”
P06: “A modern single-story house with flat roof, large windows, and an attached carport.”
P07: “A low-poly pickup truck with visible wheels, cabin, and cargo bed, game-asset style.”
P08: “A small sailing boat with hull, single mast, boom, and simplified sail.”
P09: “A potted plant: terracotta pot with a leafy monstera-style plant, about six large leaves.”
P10: “A low-poly island scene: small rocky island with two palm trees and a sandy beach.”
P11: “A spiral staircase with twelve steps winding around a central column, with a handrail.”
P12: “A chess set arrangement: checkered board with the two front rows of white pieces placed correctly.”