Benchmark Almanac

OmniCAD

Llama 4 Maverick 100% leads since August 2026
Not in model cards

Models choose parts from a component library and predict their assembly positions, orientations, and connections. The evaluator measures pose accuracy and how well connections match.

Nine multimodal models have results for five direct input configurations and three agent configurations. The authors provide sample assemblies and evaluation code.

Released
August 2026
Kind
CAD models
Level
University
Scored
Checked automatically
Answers
Agent task
Measures
Capability
Language
English
In and out
Text, Images, 3D → Actions, Text, 3D
Interaction
Agentic
Environment
APIs and tools
Items from
Real work and use
Refresh
One fixed set
Run by
Anyone
Strongest model tested
GPT-5.5 #21 on the capability index
Last model added
June 2026
Scores refreshed
November 20th, 2026
9
Entries on its own leaderboard, 8 models we track

6 models share the top score of 100% among models released by June 2026 6 models share the top score of 100% among models released by June 2026

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. The tables report no overall score. The headline is ID-NR, their first accuracy measure. 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 June 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Ministral 2DVis 100%
  2. 2 Llama 4 Maverick 2DVis 100%
  3. 3 GPT-5 mini Conflict+2DVis 100%
  4. 4 Gemini 2.5 Flash-Lite Sep 2025, 2DVis 100%
  5. 5 Gemini 3.1 Pro SameColor 100%
  6. 6 Qwen3.5-9B SameColor 100%
  7. 7 GPT-5.5 SameColor 100%
  8. 8 Claude Opus 4.8 2DVis 98.6%
  9. 9 Qwen3.7-Plus RefOBJ 98.3%

Source: Official OmniCAD leaderboard

#EntryScoreSettingDateSource
1 Ministral 14B 100% 2DVis, Agent Official leaderboard
2 Llama 4 Maverick 100% 2DVis, Agent Apr 2025 Official leaderboard
3 GPT-5 mini 100% Conflict+2DVis, Agent Aug 2025 Official leaderboard
4 Gemini 2.5 Flash-Lite Sep 2025 100% 2DVis, Agent Sep 2025 Official leaderboard
5 Gemini 3.1 Pro 100% SameColor, Direct Feb 2026 Official leaderboard
6 Qwen3.5-9B 100% SameColor, Direct Feb 2026 Official leaderboard
7 GPT-5.5 100% SameColor, Direct Apr 2026 Official leaderboard
8 Claude Opus 4.8 98.6% 2DVis, Agent May 2026 Official leaderboard
9 Qwen3.7-Plus 98.3% RefOBJ, Direct Jun 2026 Official leaderboard
#ModelScoreSettingDateSource
1 Llama 4 Maverick 100% 2DVis, Agent Apr 2025 Official leaderboard
2 GPT-5 mini 100% Conflict+2DVis, Agent Aug 2025 Official leaderboard
3 Gemini 2.5 Flash-Lite Sep 2025 100% 2DVis, Agent Sep 2025 Official leaderboard
4 Gemini 3.1 Pro 100% SameColor, Direct Feb 2026 Official leaderboard
5 Qwen3.5-9B 100% SameColor, Direct Feb 2026 Official leaderboard
6 GPT-5.5 100% SameColor, Direct Apr 2026 Official leaderboard
7 Claude Opus 4.8 98.6% 2DVis, Agent May 2026 Official leaderboard
8 Qwen3.7-Plus 98.3% RefOBJ, Direct Jun 2026 Official leaderboard

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Llama 4 Maverick Agent
Conflict+2D… 2DVis
100%
GPT-5 mini Agent
2DVis
100%
Gemini 2.5 Flash-Lite Agent
2DVis
100%
Gemini 3.1 Pro Direct
RefOBJ ColorView SameColor
100%
Gemini 3.1 Pro Agent
2DVis Conflict+2D…
100%
Qwen3.5-9B Direct
SameColor
100%
GPT-5.5 Direct
ColorView SameColor
100%
GPT-5.5 Agent
2DVis
100%
Claude Opus 4.8 Agent
2DVis
98.6%
Qwen3.7-Plus Direct
SameColor RefOBJ
98.3%
Qwen3.7-Plus Agent
Conflict+2D…
97.9%
Llama 4 Maverick Direct
ColorView RefOBJ
97.5%

Source: Official OmniCAD leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Llama 4 Maverick 2DVis Agent 100%
GPT-5 mini Conflict+2DVis Agent 100%
Gemini 2.5 Flash-Lite Sep 2025 2DVis Agent 100%
Gemini 2.5 Flash-Lite Sep 2025 Conflict+2DVis Agent 100%
Gemini 3.1 Pro SameColor Direct 100%
Gemini 3.1 Pro Conflict+2DVis Agent 100%
Qwen3.5-9B SameColor Direct 100%
Qwen3.5-9B RefOBJ Direct 100%
Qwen3.5-9B ColorView Direct 100%
GPT-5.5 SameColor Direct 100%
GPT-5.5 RefOBJ Direct 100%
GPT-5.5 Conflict+2DVis Agent 100%
Claude Opus 4.8 2DVis Agent 98.6%
Claude Opus 4.8 Conflict+2DVis Agent 98.6%
GPT-5.5 2DVis Agent 98.4%
Qwen3.7-Plus RefOBJ Direct 98.3%
GPT-5 mini 2DVis Agent 98.1%
Qwen3.7-Plus 2DVis Agent 97.9%
Qwen3.7-Plus Conflict+2DVis Agent 97.6%
Llama 4 Maverick RefOBJ Direct 97.5%
Gemini 3.1 Pro 2DVis Agent 96.8%
Gemini 3.1 Pro ColorView Direct 96.1%
Llama 4 Maverick Conflict+2DVis Agent 96%
Gemini 2.5 Flash-Lite Sep 2025 RefOBJ Direct 96%
Gemini 2.5 Flash-Lite Sep 2025 SameColor Direct 95.7%
Gemini 2.5 Flash-Lite Sep 2025 ColorView Direct 95.7%
Llama 4 Maverick SameColor Direct 95.4%
Claude Opus 4.8 RefOBJ Direct 95.3%
Claude Opus 4.8 ColorView Direct 95.1%
Claude Opus 4.8 SameColor Direct 95%
Llama 4 Maverick ColorView Direct 94.6%
Qwen3.7-Plus ColorView Direct 94.4%
Qwen3.7-Plus SameColor Direct 94.3%
GPT-5 mini SameColor Direct 93.8%
GPT-5 mini ColorView Direct 93.8%
Qwen3.5-9B 2DVis Agent 92.9%
GPT-5.5 ColorView Direct 92.9%
GPT-5 mini RefOBJ Direct 90%
Qwen3.5-9B Conflict+2DVis Agent 88.9%
Gemini 3.1 Pro RefOBJ Direct 75%

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Llama 4 Maverick Agent
Conflict+2D… 2DVis
100%
GPT-5 mini Agent
2DVis
100%
Gemini 2.5 Flash-Lite Agent
2DVis
100%
Gemini 3.1 Pro Direct
RefOBJ ColorView SameColor
100%
Gemini 3.1 Pro Agent
2DVis Conflict+2D…
100%
Qwen3.5-9B Direct
SameColor
100%
GPT-5.5 Direct
ColorView SameColor
100%
GPT-5.5 Agent
2DVis
100%
Claude Opus 4.8 Agent
2DVis
98.6%
Qwen3.7-Plus Direct
SameColor RefOBJ
98.3%
Qwen3.7-Plus Agent
Conflict+2D…
97.9%
Llama 4 Maverick Direct
ColorView RefOBJ
97.5%

Source: Official OmniCAD leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Llama 4 Maverick 2DVis Agent 100%
GPT-5 mini Conflict+2DVis Agent 100%
Gemini 2.5 Flash-Lite Sep 2025 2DVis Agent 100%
Gemini 2.5 Flash-Lite Sep 2025 Conflict+2DVis Agent 100%
Gemini 3.1 Pro SameColor Direct 100%
Gemini 3.1 Pro Conflict+2DVis Agent 100%
Qwen3.5-9B SameColor Direct 100%
Qwen3.5-9B RefOBJ Direct 100%
Qwen3.5-9B ColorView Direct 100%
GPT-5.5 SameColor Direct 100%
GPT-5.5 RefOBJ Direct 100%
GPT-5.5 Conflict+2DVis Agent 100%
Claude Opus 4.8 2DVis Agent 98.6%
Claude Opus 4.8 Conflict+2DVis Agent 98.6%
GPT-5.5 2DVis Agent 98.4%
Qwen3.7-Plus RefOBJ Direct 98.3%
GPT-5 mini 2DVis Agent 98.1%
Qwen3.7-Plus 2DVis Agent 97.9%
Qwen3.7-Plus Conflict+2DVis Agent 97.6%
Llama 4 Maverick RefOBJ Direct 97.5%
Gemini 3.1 Pro 2DVis Agent 96.8%
Gemini 3.1 Pro ColorView Direct 96.1%
Llama 4 Maverick Conflict+2DVis Agent 96%
Gemini 2.5 Flash-Lite Sep 2025 RefOBJ Direct 96%
Gemini 2.5 Flash-Lite Sep 2025 SameColor Direct 95.7%
Gemini 2.5 Flash-Lite Sep 2025 ColorView Direct 95.7%
Llama 4 Maverick SameColor Direct 95.4%
Claude Opus 4.8 RefOBJ Direct 95.3%
Claude Opus 4.8 ColorView Direct 95.1%
Claude Opus 4.8 SameColor Direct 95%
Llama 4 Maverick ColorView Direct 94.6%
Qwen3.7-Plus ColorView Direct 94.4%
Qwen3.7-Plus SameColor Direct 94.3%
GPT-5 mini SameColor Direct 93.8%
GPT-5 mini ColorView Direct 93.8%
Qwen3.5-9B 2DVis Agent 92.9%
GPT-5.5 ColorView Direct 92.9%
GPT-5 mini RefOBJ Direct 90%
Qwen3.5-9B Conflict+2DVis Agent 88.9%
Gemini 3.1 Pro RefOBJ Direct 75%

Number one over time

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

Meta
Own leaderboard Passed 85% in Aug 2026, by Llama 4 Maverick

Source: Official OmniCAD leaderboard

Llama 4 Maverick on the official leaderboard passed 85% in April 2025

Models released before it came out already scored that high. 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
0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Llama 4 Maverick 0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Llama 4 Maverick

Source: Official OmniCAD 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 Engineering and CAD

All 112 in Engineering and CAD
  • EEBench In use 13 electrical-engineering circuit and PCB design tasks graded by deterministic build checks, simulation and BOM constraints. No LLM judge. Gemini 4 Carbon 83.9%
  • 3DCodeBench In use Agentic procedural 3D modeling via code: author engine-ready 3D assets through software APIs. Graded for executability and shape fidelity. LongCat-2.5 55.4%
  • CADGenBench Only in system cards Agents construct CAD models from design prompts. The generation split is graded for executability and geometric correctness. Claude Fable 5 45.1%
  • PartBench Only in system cards Parametric mechanical part design and CAD-style generation, graded for geometric and manufacturing fidelity. Grok 4.6 58.5%
  • Razavi-Bench Not in model cards Models explain what happens in analog circuits and why. A language model awards up to four points per answer against a reference and a rubric. Gemini 4 Carbon 98.3%
  • CircuitRubric Not in model cards Models write analog circuit netlists from design requests. A program compares transistor connections and relative device sizes with accepted reference circuits. Claude Opus 4.8 96%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark

Catalog record 1 warning 1 pass
  • Warning Task access We don’t know whether the tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (9 entries).
  • Pass Models scored 8 models we track, plus 1 other entry on its leaderboard.
  • Pass Newest score The newest is from June 2026, for Qwen3.7-Plus.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is GPT-5.5 (#21).
  • Note Headroom Saturated. The best is 100%, past 85% since April 2025.
Leaderboard import 3 pass
  • Pass Importer We read the board on omnieda.com: 40 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.

How we check