Benchmark Almanac

CAD Arena

Gemini 4 Argon 0.75 leads since September 2026
Not in model cards

Agents rebuild licensed engineering drawings as editable CAD parts through the scripting APIs of NX, SolidWorks, Onshape, Fusion or Build123d. A verifier scores geometric match.

90 tasks scored by Normal's FrontierCAD verifier on a 0 to 100 scale. 12 models in the Fireworks index.

Made by
Normal Research
Released
September 2026
Kind
CAD models
Level
Expert
Scored
Checked automatically
Answers
Agent task
Tasks
Kept private
Strongest model tested
GPT-6 Astra #3 on the capability index
Last model added
September 2026
Scores refreshed
November 12th, 2026
13
Entries on its own leaderboard, 12 models we track

Gemini 4 Argon leads with 0.75

Best score per entry from the benchmark's own leaderboard, read on November 12th, 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.

13 entries. Scroll sideways for the rest.

  1. 0.75 Gemini 4 Argon
  2. 0.75 Claude Opus 5.5
  3. 0.67 GPT-6 Astra
  4. 0.66 Claude Fable 5.1
  5. 0.53 GPT-6 Sol
  6. 0.41 Gemini 3.8 Flash
  7. 0.35 Grok 4.6
  8. 0.32 Muse 1.3
  9. 0.30 GPT-6 Luna
  10. 0.26 Grok 4.7
  11. 0.19 DeepSeek-V4-Pro
  12. 0.14 Inkling-Small
  13. 0.12 Inkling
  1. 1 Gemini 4 Argon 0.75
  2. 2 Claude Opus 5.5 0.75
  3. 3 GPT-6 Astra 0.67
  4. 4 Claude Fable 5.1 0.66
  5. 5 GPT-6 Sol 0.53
  6. 6 Gemini 3.8 Flash 0.41
  7. 7 Grok 4.6 0.35
  8. 8 Muse 1.3 0.32
  9. 9 GPT-6 Luna 0.30
  10. 10 Grok 4.7 0.26
  11. 11 DeepSeek-V4-Pro 0.19
  12. 12 Inkling-Small 0.14
  13. 13 Inkling 0.12

Source: Official CAD Arena leaderboard

#EntryScoreSettingDateSource
1 Gemini 4 Argon 0.75 Gemini CLI Sep 2026 Official leaderboard
2 Claude Opus 5.5 0.75 Claude Code Sep 2026 Official leaderboard
3 GPT-6 Astra 0.67 Codex Sep 2026 Official leaderboard
4 Claude Fable 5.1 0.66 Claude Code Sep 2026 Official leaderboard
5 GPT-6 Sol 0.53 Codex Sep 2026 Official leaderboard
6 Gemini 3.8 Flash 0.41 Gemini CLI Sep 2026 Official leaderboard
7 Grok 4.6 0.35 Grok Build Aug 2026 Official leaderboard
8 Muse 1.3 0.32 Muse Code Official leaderboard
9 GPT-6 Luna 0.30 Codex Sep 2026 Official leaderboard
10 Grok 4.7 0.26 Grok Build Sep 2026 Official leaderboard
11 DeepSeek-V4-Pro 0.19 OpenCode Apr 2026 Official leaderboard
12 Inkling-Small 0.14 OpenCode Jul 2026 Official leaderboard
13 Inkling 0.12 OpenCode Jul 2026 Official leaderboard

The best score, 0.75, costs $13 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

DeepSeek OpenAI Google DeepMind Cost frontier One model at each effort level
0.00 0.20 0.40 0.60 0.80 $0.10 $1.00 $10 $100 Cost per task, log scale Gemini 4 Argon GPT-6 Astra GPT-6 Sol GPT-6 Luna DeepSeek-V4-Pro 0.00 0.20 0.40 0.60 0.80 $0.10 $1.00 $10 $100 Cost per task, log scale Gemini 4 Argon GPT-6 Astra

Source: Official CAD Arena leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 4 Argon Gemini CLI 0.75 $13
Claude Opus 5.5 Claude Code 0.75 $16
GPT-6 Astra Codex 0.67 $7.13
Claude Fable 5.1 Claude Code 0.66 $18
GPT-6 Sol Codex 0.53 $1.88
Gemini 3.8 Flash Gemini CLI 0.41 $8.83
Grok 4.6 Grok Build 0.35 $7.44
GPT-6 Luna Codex 0.30 $1.85
Grok 4.7 Grok Build 0.26 $16
DeepSeek-V4-Pro OpenCode 0.19 $0.73
Inkling-Small OpenCode 0.14 $1.36
Inkling OpenCode 0.12 $1.30

4 models have held number one on the official leaderboard since its release in September 2026

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

Anthropic OpenAI Google DeepMind
Own leaderboard 4 leaders
Gemini 4 Argon 0.75

Source: Official CAD Arena leaderboard

Scores over time

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
0.00 0.20 0.40 0.60 0.80 2026 0.00 0.20 0.40 0.60 0.80 2026

Source: Official CAD Arena leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Engineering and CAD

All 23 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%
  • Parametric CAD Bench In use Agents author editable parametric FreeCAD part designs, graded against held-back references for geometry and spec consistency. Claude Opus 5.5 61%
  • 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%
  • LMBuild Not in model cards Language model agents design 3D structures as parts, joints, materials and assembly steps. Scoring checks that the structures can be built and work as intended. GPT-6 Astra 66.5%