Benchmark Almanac

LLM 3D Worlds Arena

GPT-6 Astra 1,253 leads since September 2026
Not in model cards

Models write Blender or three.js programs that build explorable 3D scenes. Voters compare the resulting scenes in a browser without seeing the model names.

The authors rank each combination of model and rendering tool separately.

Made by
3d-arena
Kind
Building whole projects
Level
Everyday
Scored
Ranked by votes
Answers
Code
Tasks
All public
Measures
Preference
Language
English
In and out
Text → Code
Interaction
Single turn
Items from
Live play or votes
Refresh
Live
Run by
A host
In views
Kept fresh
Strongest model tested
GPT-6.2 Astra #1 on the capability index
Last model added
November 2026
Scores refreshed
November 20th, 2026
12
Models scored on its own leaderboard

GPT-6 Astra leads with 1,253

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best.

Bars start at 800.

  1. 1 GPT-6 Astra mesh/glb 1,253
  2. 2 GPT-6.2 Sol mesh/glb 1,227
  3. 3 Claude Fable 5.2 mesh/glb 1,197
  4. 4 GPT-6.1 Sol mesh/glb 1,162
  5. 5 GPT-6.2 Astra mesh/glb 1,123
  6. 6 Kimi K3.1 Pro mesh/glb 1,061
  7. 7 Claude Opus 5.5 mesh/glb 1,006
  8. 8 Inkling 2 Small mesh/glb 920
  9. 9 Claude Fable 5.1 mesh/glb 897
  10. 10 GLM-5.3 mesh/glb 888
  11. 11 Kimi K3 mesh/glb 874
  12. 12 Qwen3.8-2.4T-A95B mesh/glb 820

Source: Official LLM 3D Worlds Arena leaderboard

#ModelScore ±SettingDateLinkSource
1 GPT-6 Astra 1,253 1224.8 to 1291.4 mesh/glb, three.js Sep 2026 3d-arena-llm-3d-worlds-arena.hf.space Official leaderboard
2 GPT-6.2 Sol 1,227 1000 to 1362 mesh/glb, three.js Nov 2026 3d-arena-llm-3d-worlds-arena.hf.space Official leaderboard
3 Claude Fable 5.2 1,197 1124.4 to 1308.4 mesh/glb, Blender Nov 2026 3d-arena-llm-3d-worlds-arena.hf.space Official leaderboard
4 GPT-6.1 Sol 1,162 1132.7 to 1199.5 mesh/glb, three.js Sep 2026 3d-arena-llm-3d-worlds-arena.hf.space Official leaderboard
5 GPT-6.2 Astra 1,123 1000 to 1220.3 mesh/glb, Blender Nov 2026 3d-arena-llm-3d-worlds-arena.hf.space Official leaderboard
6 Kimi K3.1 Pro 1,061 ±76.5 mesh/glb, Blender Oct 2026 3d-arena-llm-3d-worlds-arena.hf.space Official leaderboard
7 Claude Opus 5.5 1,006 978.1 to 1046.3 mesh/glb, three.js Sep 2026 3d-arena-llm-3d-worlds-arena.hf.space Official leaderboard
8 Inkling 2 Small 920 859.3 to 1024.9 mesh/glb, three.js Nov 2026 3d-arena-llm-3d-worlds-arena.hf.space Official leaderboard
9 Claude Fable 5.1 897 872.5 to 935.3 mesh/glb, three.js Sep 2026 3d-arena-llm-3d-worlds-arena.hf.space Official leaderboard
10 GLM-5.3 888 856.5 to 917.4 mesh/glb, three.js Aug 2026 3d-arena-llm-3d-worlds-arena.hf.space Official leaderboard
11 Kimi K3 874 844.8 to 909.3 mesh/glb, three.js Jul 2026 3d-arena-llm-3d-worlds-arena.hf.space Official leaderboard
12 Qwen3.8-2.4T-A95B 820 793.3 to 852.3 mesh/glb, three.js Aug 2026 3d-arena-llm-3d-worlds-arena.hf.space Official leaderboard

4 models have held number one on the official leaderboard since 2023

Lab claims count from the day the card came out, independent runs from the model’s release.

Moonshot AI Z.ai Anthropic OpenAI
Own leaderboard 4 leaders
GPT-6 Astra 1,253

Source: Official LLM 3D Worlds Arena leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard
600 800 1,000 1,200 1,400 2027 600 800 1,000 1,200 1,400 2027

Source: Official LLM 3D Worlds Arena leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Coding

All 523 in Coding
  • SWE-bench Verified In use 500 human-validated GitHub issues from 12 Python repos. Agent writes a patch that must pass hidden tests. Claude Opus 4.7 83.5%
  • LiveCodeBench v6 In use Competitive programming problems collected after model cutoffs from LeetCode, AtCoder, Codeforces. Scored pass@1 on hidden tests. DeepSeek-V4-Pro 93.5%
  • SWE-bench Multilingual In use 300 GitHub-issue tasks across nine programming languages in SWE-bench format. Patch graded by repository tests. Claude Fable 5.2 96.6%
  • DeepSWE In use 113 original long-horizon engineering tasks across 91 repos and five languages, written from scratch. Hand-written verifiers accept alternative solutions. GPT-6 Astra 74.1%
  • SciCode In use Scientist-curated research coding problems split into subproblems across natural sciences. Code checked against test cases. Seed2.1 Pro 59.8%
  • Codeforces rating In use Model solves recent Codeforces contest problems. An Elo-style rating is computed from simulated contest performance. DeepSeek V4.1 Flash 3,471

Data audit

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (12 entries).
  • Pass Models scored 12 models we track.
  • Pass Newest score The newest is from November 2026, for GPT-6.2 Astra.
  • Pass Frontier models 6 of the top 10 have a score, led by GPT-6.2 Astra (#1).
  • Note Headroom Scored as a rating, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on 3d-arena-llm-3d-worlds-arena.hf.space: 24 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check