Benchmark Almanac

Parametric CAD Bench

Claude Opus 5.5 61% leads since Sep 2026 In use

Agents author editable parametric FreeCAD part designs, graded against held-back references for geometry and spec consistency.

Not the same as anniedoris/CADBench. A submission-portal repo exists at github.com/gNucleus-AI/cad-bench-submission.

Made by gNucleus AI Released 2026 CAD models Tasks kept private Last in a launch post Aug 2026 Last model added Sep 2026 Scores refreshed 2 Nov 2026
Level
Expert Specialists with years of training, often over hours of work
Scored
Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
Answers
Agent task Many steps in a terminal, browser, desktop, app, game or robot.
Leaderboard
11
Models scored on its own leaderboard
1
Launch post lists it, from 1 lab
2
System cards report it, from 1 lab

Claude Opus 5.5 leads with 61%

Best score per model from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best.

  1. 61.0 Claude Opus 5.5
  2. 56.9 GPT-6 Astra
  3. 56.8 Claude Fable 5.1
  4. 50.9 Claude Opus 5
  5. 36.2 Gemini 3.8 Flash
  6. 34.8 Grok 4.7
  7. 30.0 GPT-5.6 Sol
  8. 27.7 Grok 4.6
  9. 24.9 Kimi K3
  10. 20.9 Muse Spark 1.3
  11. 17.2 GPT-5.6 Terra
  1. 1 Claude Opus 5.5 max 61%
  2. 2 GPT-6 Astra max 56.9%
  3. 3 Claude Fable 5.1 max 56.8%
  4. 4 Claude Opus 5 max 50.9%
  5. 5 Gemini 3.8 Flash high 36.2%
  6. 6 Grok 4.7 high 34.8%
  7. 7 GPT-5.6 Sol max 30%
  8. 8 Grok 4.6 high 27.7%
  9. 9 Kimi K3 max 24.9%
  10. 10 Muse Spark 1.3 max 20.9%
  11. 11 GPT-5.6 Terra max 17.2%
#ModelScoreSettingDateSource
1 Claude Opus 5.5 61% max, Claude Code Sep 2026 Official leaderboard
2 GPT-6 Astra 56.9% max, Codex Sep 2026 Official leaderboard
3 Claude Fable 5.1 56.8% max, Claude Code Sep 2026 Official leaderboard
4 Claude Opus 5 50.9% max, Claude Code Jul 2026 Official leaderboard
5 Gemini 3.8 Flash 36.2% high, mini-swe-agent Sep 2026 Official leaderboard
6 Grok 4.7 34.8% high, Grok Build Sep 2026 Official leaderboard
7 GPT-5.6 Sol 30% max, Codex Jul 2026 Official leaderboard
8 Grok 4.6 27.7% high, Grok Build Aug 2026 Official leaderboard
9 Kimi K3 24.9% max, mini-swe-agent Jul 2026 Official leaderboard
10 Muse Spark 1.3 20.9% max, mini-swe-agent Sep 2026 Official leaderboard
11 GPT-5.6 Terra 17.2% max, Codex Jul 2026 Official leaderboard

Labs report scores for 3 models in their cards. We plot every source in the charts below.

5 models have held number one since its release in Jan 2026

Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard and what labs claim in their cards.

OpenAI Anthropic xAI Qwen
Own leaderboard 5 leaders
Claude Opus 5.5 61%
Lab claims Passed 85% in Jul 2026, by Grok 4.5
Qwen 3.8 Max 91.5%

The best score passed 85% in Jul 2026

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

On the benchmark’s own leaderboard Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% Near the ceiling (85%) Qwen 3.8 Max GPT-5.6 Sol 0% 20% 40% 60% 80% 100% Near the ceiling (85%) Qwen 3.8 Max GPT-5.6 Sol

One lab put it in a launch post in the last six months

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

Qwen
DateLaunch postScoreAs the post writes it
3 Aug 2026 Qwen 3.8 Max 91.5 Parametric CAD Bench

2 system cards report it, from 1 lab

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
12 Aug 2026 Grok 4.6 model card 88.4 CADBench (Parametric CAD Bench), reward, xhigh, Grok Build harness, run by gNucleus (4.5). High = 87.8% Capabilities
14 Jul 2026 Grok 4.5 model card 88.1 CAD-Bench (Parametric CAD Bench), reward, high (3.3) Capabilities

More in Engineering and CAD

All 18 in Engineering and CAD
  • BenchCAD In use Programmatic CAD: 17,900 execution-verified CadQuery programs over 106 industrial part families. Four matched tasks scored on geometry. GPT-6 Astra 95.9%
  • EEBench In use 13 electrical-engineering circuit and PCB design tasks graded by deterministic build checks, simulation and BOM constraints. No LLM judge. Claude Opus 5.5 75%
  • 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%
  • 3DCodeBench Only in system cards Agentic procedural 3D modeling via code: author engine-ready 3D assets through software APIs. Graded for executability and shape fidelity. Grok 4.6 54%
  • PartBench Only in system cards Parametric mechanical part design and CAD-style generation, graded for geometric and manufacturing fidelity. Grok 4.6 58.5%
  • Physical-Engineering's Last Exam Not in model cards Agents produce billable engineering deliverables such as calculation packages, drawings and design reviews in 85 tasks across electrical, civil, structural and mechanical engineering. Automatic graders check constraints, safety limits and the design objective. Claude Opus 5.5 0.49