UnityBench-Mini3D
Agents modify existing Unity projects. EditMode and PlayMode tests check the saved scene, component settings, and runtime behavior, with partial credit for completed milestones.
Unity publishes 36 configurations covering 21 models on 77 tasks. Its repository shows reference scenes and lists task projects and grading software as forthcoming.
- Made by
- Unity China AI Hub and collaborators
- Released
- October 2026
- Kind
- Building whole projects
- Level
- University
- Scored
- Checked automatically
- Answers
- Agent task
- Tasks
- Kept private
- Measures
- Capability
- Language
- English
- Interaction
- Agentic
- Environment
- Unity
- Refresh
- One fixed set
- Run by
- A host
- In views
- Agents
- Strongest model tested
- Claude Opus 5 #11 on the capability index
- Last model added
- August 2026
- Scores refreshed
- November 20th, 2026
GPT-5.6 Sol leads with 72.5% among models released by August 2026 GPT-5.6 Sol leads with 72.5% among models released by August 2026
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in August 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 13 more
- 72.5 GPT-5.6 Sol
- 71.8 Claude Opus 5
- 71.2 Grok 4.5
- 70.6 Grok 4.6
- 70.5
DeepSeek-V4-Pro
- 68.0 GPT-5.6 Terra
- 66.7
Qwen 3.8 Max
- 66.4 GPT-5.6 Luna
- 66.3 Kimi K3
- 66.2 GPT-5.5
- 65.0
DeepSeek-V4-Flash
- 62.5 GLM-5.2
- 54.0 Claude Sonnet 5
- 52.3
MiniMax-M3
- 51.0 Hunyuan 3
- 47.8
Qwen3.7-Max
- 41.5 Kimi K2.7 Code
- 38.9
Qwen3.7-Plus
- 31.2
Gemini 3.7 Flash
- 25.2
Gemini 3.1 Pro
- 24.4
Seed2.1 Pro
- 1 GPT-5.6 Sol 72.5%
- 2 Claude Opus 5 71.8%
- 3 Grok 4.5 71.2%
- 4 Grok 4.6 70.6%
- 5
DeepSeek-V4-Pro 70.5%
- 6 GPT-5.6 Terra 68%
- 7
Qwen 3.8 Max 66.7%
- 8 GPT-5.6 Luna 66.4%
- 9 Kimi K3 66.3%
- 10 GPT-5.5 66.2%
- 11
DeepSeek-V4-Flash 65%
- 12 GLM-5.2 62.5%
- 13 Claude Sonnet 5 54%
- 14
MiniMax-M3 52.3%
- 15 Hunyuan 3 51%
- 16
Qwen3.7-Max 47.8%
- 17 Kimi K2.7 Code 41.5%
- 18
Qwen3.7-Plus 38.9%
- 19
Gemini 3.7 Flash 31.2%
- 20
Gemini 3.1 Pro 25.2%
- 21
Seed2.1 Pro 24.4%
Source: Official UnityBench-Mini3D leaderboard
- 72.5 GPT-5.6 Sol
- 71.8 Claude Opus 5
- 71.2 Grok 4.5
- 70.6 Grok 4.6
- 70.5
DeepSeek-V4-Pro
- 68.0 GPT-5.6 Terra
- 66.7
Qwen 3.8 Max
- 66.4 GPT-5.6 Luna
- 66.3 Kimi K3
- 66.2 GPT-5.5
- 65.0
DeepSeek-V4-Flash
- 62.5 GLM-5.2
- 54.0 Claude Sonnet 5
- 52.3
MiniMax-M3
- 47.8
Qwen3.7-Max
- 41.5 Kimi K2.7 Code
- 38.9
Qwen3.7-Plus
- 31.2
Gemini 3.7 Flash
- 25.2
Gemini 3.1 Pro
- 24.4
Seed2.1 Pro
- 1 GPT-5.6 Sol 72.5%
- 2 Claude Opus 5 71.8%
- 3 Grok 4.5 71.2%
- 4 Grok 4.6 70.6%
- 5
DeepSeek-V4-Pro 70.5%
- 6 GPT-5.6 Terra 68%
- 7
Qwen 3.8 Max 66.7%
- 8 GPT-5.6 Luna 66.4%
- 9 Kimi K3 66.3%
- 10 GPT-5.5 66.2%
- 11
DeepSeek-V4-Flash 65%
- 12 GLM-5.2 62.5%
- 13 Claude Sonnet 5 54%
- 14
MiniMax-M3 52.3%
- 15
Qwen3.7-Max 47.8%
- 16 Kimi K2.7 Code 41.5%
- 17
Qwen3.7-Plus 38.9%
- 18
Gemini 3.7 Flash 31.2%
- 19
Gemini 3.1 Pro 25.2%
- 20
Seed2.1 Pro 24.4%
Source: Official UnityBench-Mini3D leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.6 Sol | 72.5% | Codex | Jul 2026 | Official leaderboard |
| 2 | Claude Opus 5 | 71.8% | Claude Code | Jul 2026 | Official leaderboard |
| 3 | Grok 4.5 | 71.2% | Grok Build | Jul 2026 | Official leaderboard |
| 4 | Grok 4.6 | 70.6% | Grok Build | Aug 2026 | Official leaderboard |
| 5 | | 70.5% | DS Harness | Apr 2026 | Official leaderboard |
| 6 | GPT-5.6 Terra | 68% | Codex | Jul 2026 | Official leaderboard |
| 7 | | 66.7% | Codely | Aug 2026 | Official leaderboard |
| 8 | GPT-5.6 Luna | 66.4% | Codex | Jul 2026 | Official leaderboard |
| 9 | Kimi K3 | 66.3% | Kimi Code | Jul 2026 | Official leaderboard |
| 10 | GPT-5.5 | 66.2% | Codex | Apr 2026 | Official leaderboard |
| 11 | | 65% | DS Harness | Apr 2026 | Official leaderboard |
| 12 | GLM-5.2 | 62.5% | Codely | Jun 2026 | Official leaderboard |
| 13 | Claude Sonnet 5 | 54% | Claude Code | Jun 2026 | Official leaderboard |
| 14 | | 52.3% | Codely | Jun 2026 | Official leaderboard |
| 15 | Hunyuan 3 | 51% | Codely | Official leaderboard | |
| 16 | | 47.8% | Qwen Code | May 2026 | Official leaderboard |
| 17 | Kimi K2.7 Code | 41.5% | Kimi Code | Jun 2026 | Official leaderboard |
| 18 | | 38.9% | Qwen Code | Jun 2026 | Official leaderboard |
| 19 | | 31.2% | Antigravity | Aug 2026 | Official leaderboard |
| 20 | | 25.2% | Antigravity | Feb 2026 | Official leaderboard |
| 21 | | 24.4% | Codely | Jun 2026 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.6 Sol | 72.5% | Codex | Jul 2026 | Official leaderboard |
| 2 | Claude Opus 5 | 71.8% | Claude Code | Jul 2026 | Official leaderboard |
| 3 | Grok 4.5 | 71.2% | Grok Build | Jul 2026 | Official leaderboard |
| 4 | Grok 4.6 | 70.6% | Grok Build | Aug 2026 | Official leaderboard |
| 5 | | 70.5% | DS Harness | Apr 2026 | Official leaderboard |
| 6 | GPT-5.6 Terra | 68% | Codex | Jul 2026 | Official leaderboard |
| 7 | | 66.7% | Codely | Aug 2026 | Official leaderboard |
| 8 | GPT-5.6 Luna | 66.4% | Codex | Jul 2026 | Official leaderboard |
| 9 | Kimi K3 | 66.3% | Kimi Code | Jul 2026 | Official leaderboard |
| 10 | GPT-5.5 | 66.2% | Codex | Apr 2026 | Official leaderboard |
| 11 | | 65% | DS Harness | Apr 2026 | Official leaderboard |
| 12 | GLM-5.2 | 62.5% | Codely | Jun 2026 | Official leaderboard |
| 13 | Claude Sonnet 5 | 54% | Claude Code | Jun 2026 | Official leaderboard |
| 14 | | 52.3% | Codely | Jun 2026 | Official leaderboard |
| 15 | | 47.8% | Qwen Code | May 2026 | Official leaderboard |
| 16 | Kimi K2.7 Code | 41.5% | Kimi Code | Jun 2026 | Official leaderboard |
| 17 | | 38.9% | Qwen Code | Jun 2026 | Official leaderboard |
| 18 | | 31.2% | Antigravity | Aug 2026 | Official leaderboard |
| 19 | | 25.2% | Antigravity | Feb 2026 | Official leaderboard |
| 20 | | 24.4% | Codely | Jun 2026 | Official leaderboard |
Rankings by part
We read 10 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as User interface, Animation and Editor operations. Claude Opus 5 leads 3 of them, and 5 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
User interface
20 models we track have a score here.
- 1
Qwen 3.8 Max ↑6 74.3%
- 2 GPT-5.6 Terra ↑4 72.8%
- 3 Claude Opus 5 ↓1 71.1%
- 4 GPT-5.5 ↑6 67.7%
- 5 Grok 4.6 ↓1 66.1%
- 6 GPT-5.6 Sol ↓5 65.8%
- 7 Kimi K3 ↑2 65.6%
- 8
DeepSeek-V4-Pro ↓3 64.4%
- 9 GLM-5.2 ↑3 63.9%
- 10
DeepSeek-V4-Flash ↑1 61.3%
Animation
20 models we track have a score here.
- 1 Grok 4.5 ↑2 81.1%
- 2 Grok 4.6 ↑2 78.8%
- 3 Kimi K3 ↑6 78.3%
- 4 GPT-5.6 Luna ↑4 77.4%
- 5 GPT-5.6 Terra ↑1 76.4%
- 6
DeepSeek-V4-Flash ↑5 76.2%
- 7 GPT-5.5 ↑3 76%
- 8 GPT-5.6 Sol ↓7 74.7%
- 9
DeepSeek-V4-Pro ↓4 70.8%
- 10 Claude Opus 5 ↓8 69.7%
Editor operations
20 models we track have a score here.
- 1 Claude Opus 5 ↑1 88.5%
- 2 GPT-5.6 Luna ↑6 85%
- 3 GPT-5.6 Terra ↑3 83.2%
- 4 Grok 4.5 ↓1 80.8%
- 5 Grok 4.6 ↓1 79.5%
- 6
Qwen 3.8 Max ↑1 78.4%
- 7 GLM-5.2 ↑5 77.4%
- 8 GPT-5.5 ↑2 76.6%
- 9
DeepSeek-V4-Pro ↓4 75.5%
- 10 GPT-5.6 Sol ↓9 75%
Scene construction
20 models we track have a score here.
- 1 Claude Opus 5 ↑1 67.1%
- 2 GPT-5.6 Sol ↓1 61.7%
- 3 Grok 4.6 ↑1 60.5%
- 4
DeepSeek-V4-Pro ↑1 58.4%
- 5 Grok 4.5 ↓2 58.1%
- 6
Qwen 3.8 Max ↑1 56.9%
- 7 Kimi K3 ↑2 56.4%
- 8
DeepSeek-V4-Flash ↑3 52.7%
- 9 GPT-5.6 Terra ↓3 49.8%
- 10 GPT-5.5 48.3%
Particles & shaders
20 models we track have a score here.
- 1 GPT-5.6 Sol 90%
- 2
DeepSeek-V4-Pro ↑3 87.4%
- 3 Claude Opus 5 ↓1 87%
- 4 Grok 4.6 83%
- 5 Kimi K3 ↑4 82.5%
- 6
Qwen 3.8 Max ↑1 81.1%
- 7
DeepSeek-V4-Flash ↑4 80.6%
- 8 Grok 4.5 ↓5 79.9%
- 9 GPT-5.5 ↑1 78.3%
- 10 GPT-5.6 Terra ↓4 76.5%
Physics
20 models we track have a score here.
- 1 Grok 4.5 ↑2 84.6%
- 2
DeepSeek-V4-Pro ↑3 79.1%
- 3 GPT-5.6 Sol ↓2 78.7%
- 4 Grok 4.6 75.4%
- 5 Claude Opus 5 ↓3 74.1%
- 6 Kimi K3 ↑3 73.6%
- 7
DeepSeek-V4-Flash ↑4 73.2%
- 8 GLM-5.2 ↑4 71%
- 9 GPT-5.6 Luna ↓1 67.5%
- 10
Qwen 3.8 Max ↓3 67.2%
Miscellaneous
20 models we track have a score here.
- 1
DeepSeek-V4-Flash ↑10 68.8%
- 2 Claude Opus 5 62.2%
- 3 Grok 4.5 62.1%
- 4 Grok 4.6 62.1%
- 5 GPT-5.6 Sol ↓4 58.8%
- 6 Kimi K3 ↑3 58.8%
- 7 GPT-5.6 Luna ↑1 58.4%
- 8 GPT-5.5 ↑2 55.6%
- 9
DeepSeek-V4-Pro ↓4 55.1%
- 10 GLM-5.2 ↑2 55.1%
Easy
20 models we track have a score here.
- 1 Grok 4.6 ↑3 94.6%
- 2 Claude Opus 5 94.5%
- 3
Qwen 3.8 Max ↑4 93.7%
- 4 Grok 4.5 ↓1 91.5%
- 5 GPT-5.5 ↑5 91.2%
- 6 GPT-5.6 Terra 90.6%
- 7
DeepSeek-V4-Pro ↓2 90.5%
- 8 GPT-5.6 Sol ↓7 90.3%
- 9 Kimi K3 89.8%
- 10
DeepSeek-V4-Flash ↑1 89.2%
Medium
20 models we track have a score here.
- 1 Claude Opus 5 ↑1 74.3%
- 2 Grok 4.5 ↑1 70.9%
- 3
DeepSeek-V4-Pro ↑2 70.9%
- 4 Kimi K3 ↑5 68.6%
- 5 GPT-5.6 Sol ↓4 67%
- 6 Grok 4.6 ↓2 66.1%
- 7 GPT-5.6 Luna ↑1 66%
- 8 GPT-5.6 Terra ↓2 65.4%
- 9 GLM-5.2 ↑3 63.4%
- 10
DeepSeek-V4-Flash ↑1 63.3%
Hard
20 models we track have a score here.
- 1 GPT-5.6 Sol 60.1%
- 2 Claude Opus 5 54%
- 3 Grok 4.6 ↑1 53.6%
- 4 Grok 4.5 ↓1 49.4%
- 5
DeepSeek-V4-Pro 48.3%
- 6
Qwen 3.8 Max ↑1 48.3%
- 7 GPT-5.6 Terra ↓1 46.8%
- 8
DeepSeek-V4-Flash ↑3 46.2%
- 9 GPT-5.5 ↑1 42.9%
- 10 GPT-5.6 Luna ↓2 42.9%
The best score, 72.5%, costs $25 a task
Runs on the cost frontier score higher than any run that's cheaper. Grok 4.5 comes within 1.3 points of the best for $7.99 a task, 3 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.
Source: Official UnityBench-Mini3D leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-5.6 Sol | Codex | 72.5% | $25 | ||
| Claude Opus 5 | Claude Code | 71.8% | $50 | ||
| Grok 4.5 | Grok Build | 71.2% | $7.99 | ||
| Grok 4.6 | Grok Build | 70.6% | $12 | ||
| DS Harness | 70.5% | $10 | |||
| Claude Opus 5 | Codely | 68.8% | $50 | ||
| GPT-5.6 Terra | Codex | 68% | $14 | ||
| Codely | 66.7% | $13 | |||
| GPT-5.6 Luna | Codex | 66.4% | $1.94 | ||
| Kimi K3 | Kimi Code | 66.3% | $11 | ||
| Kimi K3 | Codely | 66.2% | $17 | ||
| GPT-5.5 | Codex | 66.2% | $23 | ||
| DS Harness | 65% | $4.98 | |||
| Grok 4.6 | Codely | 63.9% | $7.08 | ||
| Codely | 63.1% | $8.25 | |||
| GLM-5.2 | Codely | 62.5% | $12 | ||
| GPT-5.6 Sol | Codely | 62.3% | $19 | ||
| Qwen Code | 62.3% | $13 | |||
| Kimi K3 | OpenCode | 62% | $10 | ||
| Codely | 61.7% | $4.03 | |||
| GPT-5.6 Sol | OpenCode | 61.3% | $20 | ||
| Claude Sonnet 5 | Claude Code | 54% | $17 | ||
| Codely | 52.3% | $5.28 | |||
| GLM-5.2 | OpenCode | 52% | $7.49 | ||
| Qwen Code | 47.8% | $5.74 | |||
| Kimi K2.7 Code | Kimi Code | 41.5% | $4.29 | ||
| Qwen Code | 38.9% | $1.16 | |||
| OpenCode | 38.7% | $1.09 | |||
| Antigravity | 31.2% | $6.48 | |||
| Antigravity | 25.2% | $1.81 | |||
| Codely | 24.4% | $1.44 | |||
| GPT-5.6 Sol | UI-TARS | 23.1% | $3.74 | ||
| GPT-5.6 Sol | Agent S | 22.2% | |||
| Kimi K3 | UI-TARS | 8.77% | $2.02 | ||
| Kimi K3 | Agent S | 5.04% |
The best score, 72.5%, costs $25 a task
Runs on the cost frontier score higher than any run that's cheaper. Grok 4.5 comes within 1.3 points of the best for $7.99 a task, 3 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.
Source: Official UnityBench-Mini3D leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-5.6 Sol | Codex | 72.5% | $25 | ||
| Claude Opus 5 | Claude Code | 71.8% | $50 | ||
| Grok 4.5 | Grok Build | 71.2% | $7.99 | ||
| Grok 4.6 | Grok Build | 70.6% | $12 | ||
| DS Harness | 70.5% | $10 | |||
| Claude Opus 5 | Codely | 68.8% | $50 | ||
| GPT-5.6 Terra | Codex | 68% | $14 | ||
| Codely | 66.7% | $13 | |||
| GPT-5.6 Luna | Codex | 66.4% | $1.94 | ||
| Kimi K3 | Kimi Code | 66.3% | $11 | ||
| Kimi K3 | Codely | 66.2% | $17 | ||
| GPT-5.5 | Codex | 66.2% | $23 | ||
| DS Harness | 65% | $4.98 | |||
| Grok 4.6 | Codely | 63.9% | $7.08 | ||
| Codely | 63.1% | $8.25 | |||
| GLM-5.2 | Codely | 62.5% | $12 | ||
| GPT-5.6 Sol | Codely | 62.3% | $19 | ||
| Qwen Code | 62.3% | $13 | |||
| Kimi K3 | OpenCode | 62% | $10 | ||
| Codely | 61.7% | $4.03 | |||
| GPT-5.6 Sol | OpenCode | 61.3% | $20 | ||
| Claude Sonnet 5 | Claude Code | 54% | $17 | ||
| Codely | 52.3% | $5.28 | |||
| GLM-5.2 | OpenCode | 52% | $7.49 | ||
| Qwen Code | 47.8% | $5.74 | |||
| Kimi K2.7 Code | Kimi Code | 41.5% | $4.29 | ||
| Qwen Code | 38.9% | $1.16 | |||
| OpenCode | 38.7% | $1.09 | |||
| Antigravity | 31.2% | $6.48 | |||
| Antigravity | 25.2% | $1.81 | |||
| Codely | 24.4% | $1.44 | |||
| GPT-5.6 Sol | UI-TARS | 23.1% | $3.74 | ||
| GPT-5.6 Sol | Agent S | 22.2% | |||
| Kimi K3 | UI-TARS | 8.77% | $2.02 | ||
| Kimi K3 | Agent S | 5.04% |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official UnityBench-Mini3D leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official UnityBench-Mini3D 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
- 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 The tasks are private.
- Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
- Pass Independent runs Scores from the official leaderboard (21 entries).
- Pass Models scored 20 models we track, plus 1 other entry on its leaderboard.
- Pass Newest score The newest is from August 2026, for Gemini 3.7 Flash.
- Note Frontier models None of the top 10 has a score. The strongest tested is Claude Opus 5 (#11).
- Pass Headroom The best is 72.5%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on unitybench.github.io: 35 rows for models we track and 1 other entry 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.