PDAGENT-BENCH
Models answer chip design questions, diagnose failures, interpret reports, and write tool commands. Rules grade five categories and model judges grade three.
The authors rank 16 models on 343 questions across two rounds with different judges. Three models have scores for six of the eight categories.
- Made by
- Qiufeng Li and colleagues
- Released
- June 2026
- Kind
- Electronics
- Level
- Expert
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- Public, plus a private test set
- Measures
- Capability
- Language
- English
- Interaction
- Single turn
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- A host
- Strongest model tested
- Claude Opus 5 #11 on the capability index
- Last model added
- September 2026
- Scores refreshed
- November 20th, 2026
GPT-5.5 leads with 65.1% among models released by September 2026 GPT-5.5 leads with 65.1% among models released by September 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. 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 September 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
- Claude Opus 5.5
- and 14 more
- 1 GPT-5.5 r1, 8 categories 65.1%
- 2 GPT-5.6 Sol r2, 8 categories 63.4%
- 3 Claude Opus 5 r2, 8 categories 60.5%
- 4 Claude Opus 4.7 r1, 8 categories 59.9%
- 5
DeepSeek V4.1 Flash r2, 8 categories 57.6%
- 6 Claude Sonnet 4.6 r1, 8 categories 57.3%
- 7 GPT-5.6 Luna r2, 8 categories 55.3%
- 8 Claude Sonnet 5 r2, 8 categories 51.5%
- 9 GPT-5.4 Mini r1, 8 categories 50.2%
- 10 GPT-5.1 r1, 8 categories 50.2%
- 11 DeepSeek-Reasoner r1, 6 categories 42.5%
- 12 DeepSeek-Chat r1, 6 categories 42%
- 13 GPT-5 mini r1, 8 categories 41%
- 14 Qwen-VL-30B r1, 8 categories 31.4%
- 15 Qwen-VL-8B r1, 8 categories 27.7%
- 16
Qwen3-8B r1, 6 categories 21.2%
Source: Official PDAGENT-BENCH leaderboard
- 1 GPT-5.5 r1, 8 categories 65.1%
- 2 GPT-5.6 Sol r2, 8 categories 63.4%
- 3 Claude Opus 5 r2, 8 categories 60.5%
- 4 Claude Opus 4.7 r1, 8 categories 59.9%
- 5
DeepSeek V4.1 Flash r2, 8 categories 57.6%
- 6 Claude Sonnet 4.6 r1, 8 categories 57.3%
- 7 GPT-5.6 Luna r2, 8 categories 55.3%
- 8 Claude Sonnet 5 r2, 8 categories 51.5%
- 9 GPT-5.4 Mini r1, 8 categories 50.2%
- 10 GPT-5.1 r1, 8 categories 50.2%
- 11 GPT-5 mini r1, 8 categories 41%
- 12
Qwen3-8B r1, 6 categories 21.2%
Source: Official PDAGENT-BENCH leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.5 | 65.1% | r1, 8 categories | Apr 2026 | Official leaderboard |
| 2 | GPT-5.6 Sol | 63.4% | r2, 8 categories | Jul 2026 | Official leaderboard |
| 3 | Claude Opus 5 | 60.5% | r2, 8 categories | Jul 2026 | Official leaderboard |
| 4 | Claude Opus 4.7 | 59.9% | r1, 8 categories | Apr 2026 | Official leaderboard |
| 5 | | 57.6% | r2, 8 categories | Sep 2026 | Official leaderboard |
| 6 | Claude Sonnet 4.6 | 57.3% | r1, 8 categories | Feb 2026 | Official leaderboard |
| 7 | GPT-5.6 Luna | 55.3% | r2, 8 categories | Jul 2026 | Official leaderboard |
| 8 | Claude Sonnet 5 | 51.5% | r2, 8 categories | Jun 2026 | Official leaderboard |
| 9 | GPT-5.4 Mini | 50.2% | r1, 8 categories | Mar 2026 | Official leaderboard |
| 10 | GPT-5.1 | 50.2% | r1, 8 categories | Nov 2025 | Official leaderboard |
| 11 | DeepSeek-Reasoner open weights | 42.5% | r1, 6 categories | Official leaderboard | |
| 12 | DeepSeek-Chat open weights | 42% | r1, 6 categories | Official leaderboard | |
| 13 | GPT-5 mini | 41% | r1, 8 categories | Aug 2025 | Official leaderboard |
| 14 | Qwen-VL-30B open weights | 31.4% | r1, 8 categories | Official leaderboard | |
| 15 | Qwen-VL-8B open weights | 27.7% | r1, 8 categories | Official leaderboard | |
| 16 | | 21.2% | r1, 6 categories | Apr 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.5 | 65.1% | r1, 8 categories | Apr 2026 | Official leaderboard |
| 2 | GPT-5.6 Sol | 63.4% | r2, 8 categories | Jul 2026 | Official leaderboard |
| 3 | Claude Opus 5 | 60.5% | r2, 8 categories | Jul 2026 | Official leaderboard |
| 4 | Claude Opus 4.7 | 59.9% | r1, 8 categories | Apr 2026 | Official leaderboard |
| 5 | | 57.6% | r2, 8 categories | Sep 2026 | Official leaderboard |
| 6 | Claude Sonnet 4.6 | 57.3% | r1, 8 categories | Feb 2026 | Official leaderboard |
| 7 | GPT-5.6 Luna | 55.3% | r2, 8 categories | Jul 2026 | Official leaderboard |
| 8 | Claude Sonnet 5 | 51.5% | r2, 8 categories | Jun 2026 | Official leaderboard |
| 9 | GPT-5.4 Mini | 50.2% | r1, 8 categories | Mar 2026 | Official leaderboard |
| 10 | GPT-5.1 | 50.2% | r1, 8 categories | Nov 2025 | Official leaderboard |
| 11 | GPT-5 mini | 41% | r1, 8 categories | Aug 2025 | Official leaderboard |
| 12 | | 21.2% | r1, 6 categories | Apr 2025 | Official leaderboard |
Rankings by part
We read 23 parts from the benchmark’s own leaderboard in 2 groups: understanding and script generation. Each ranks models on one part of the test. GPT-5.5 leads 8 of them, and 7 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
Foundational Knowledge pass@1
12 models we track have a score here.
- 1 Claude Opus 5 ↑2 90%
- 2 GPT-5.6 Sol 85.7%
- 3 GPT-5.6 Luna ↑4 84.8%
- 4
DeepSeek V4.1 Flash ↑1 84.8%
- 5 Claude Sonnet 5 ↑3 82.9%
- 6 Claude Sonnet 4.6 82.6%
- 7 Claude Opus 4.7 ↓3 82.4%
- 8 GPT-5.1 ↑2 82%
- 9 GPT-5.5 ↓8 80.4%
- 10 GPT-5 mini ↑1 80%
Foundational Knowledge pass@5
12 models we track have a score here.
- 1 Claude Opus 5 ↑2 90.8%
- 2 GPT-5.6 Sol 87.1%
- 3
DeepSeek V4.1 Flash ↑2 86.3%
- 4 Claude Opus 4.7 84.2%
- 5 GPT-5.1 ↑5 84%
- 6 Claude Sonnet 4.6 83.5%
- 7 Claude Sonnet 5 ↑1 83.2%
- 8 GPT-5.5 ↓7 83%
- 9 GPT-5.6 Luna ↓2 82.9%
- 10 GPT-5 mini ↑1 82%
Foundational Knowledge human pass@1
7 models we track have a score here.
- 1 GPT-5.1 ↑9 80.2%
- 2 Claude Sonnet 4.6 ↑4 79.4%
- 3 Claude Opus 4.7 ↑1 78.4%
- 4 GPT-5.5 ↓3 77.8%
- 5 GPT-5 mini ↑6 70.1%
- 6 GPT-5.4 Mini ↑3 66.8%
- 7
Qwen3-8B ↑5 48.8%
Foundational Knowledge human pass@5
7 models we track have a score here.
- 1 GPT-5.1 ↑9 83%
- 2 Claude Opus 4.7 ↑2 82.9%
- 3 Claude Sonnet 4.6 ↑3 82.1%
- 4 GPT-5.5 ↓3 80.7%
- 5 GPT-5 mini ↑6 79%
- 6 GPT-5.4 Mini ↑3 72.6%
- 7
Qwen3-8B ↑5 57.3%
Root-Cause Analysis pass@1
12 models we track have a score here.
- 1 Claude Opus 4.7 ↑3 76.2%
- 2 Claude Opus 5 ↑1 74.8%
- 3 GPT-5.5 ↓2 73.3%
- 4 Claude Sonnet 4.6 ↑2 68.8%
- 5
DeepSeek V4.1 Flash 66.9%
- 6 GPT-5.6 Sol ↓4 65.7%
- 7 GPT-5.1 ↑3 65.5%
- 8 GPT-5.6 Luna ↓1 57.6%
- 9 Claude Sonnet 5 ↓1 57.4%
- 10 GPT-5 mini ↑1 56%
Root-Cause Analysis pass@5
12 models we track have a score here.
- 1 Claude Opus 5 ↑2 78.5%
- 2 GPT-5.5 ↓1 77.4%
- 3 Claude Opus 4.7 ↑1 76.7%
- 4 Claude Sonnet 4.6 ↑2 70%
- 5 GPT-5.1 ↑5 67.4%
- 6 GPT-5.6 Sol ↓4 67.1%
- 7
DeepSeek V4.1 Flash ↓2 65.9%
- 8 GPT-5 mini ↑3 65.5%
- 9 GPT-5.6 Luna ↓2 59%
- 10 GPT-5.4 Mini ↓1 58.1%
Root-Cause Analysis human pass@1
7 models we track have a score here.
- 1 Claude Opus 4.7 ↑3 71.4%
- 2 GPT-5.5 ↓1 68.9%
- 3 Claude Sonnet 4.6 ↑3 66%
- 4 GPT-5.1 ↑6 61.9%
- 5 GPT-5 mini ↑6 54%
- 6 GPT-5.4 Mini ↑3 50.2%
- 7
Qwen3-8B ↑5 21.6%
Root-Cause Analysis human pass@5
7 models we track have a score here.
- 1 GPT-5.5 73.2%
- 2 Claude Opus 4.7 ↑2 73.1%
- 3 Claude Sonnet 4.6 ↑3 66.4%
- 4 GPT-5.1 ↑6 62.4%
- 5 GPT-5 mini ↑6 61.5%
- 6 GPT-5.4 Mini ↑3 52.6%
- 7
Qwen3-8B ↑5 26.3%
Report Comprehension pass@1
11 models we track have a score here.
- 1 GPT-5.5 74.6%
- 2 GPT-5.6 Sol 68.2%
- 3 Claude Sonnet 4.6 ↑3 66.8%
- 4 GPT-5 mini ↑7 65.5%
- 5 GPT-5.1 ↑5 65.5%
- 6 Claude Opus 4.7 ↓2 64.1%
- 7 Claude Sonnet 5 ↑1 62.3%
- 8 GPT-5.6 Luna ↓1 60%
- 9 Claude Opus 5 ↓6 53.6%
- 10
DeepSeek V4.1 Flash ↓5 50.5%
Report Comprehension pass@5
11 models we track have a score here.
- 1 GPT-5.5 75%
- 2 Claude Sonnet 4.6 ↑4 73.6%
- 3 GPT-5.1 ↑7 70.5%
- 4 GPT-5 mini ↑7 68.5%
- 5 Claude Opus 5 ↓2 68.4%
- 6 Claude Opus 4.7 ↓2 65.9%
- 7 GPT-5.6 Sol ↓5 65.8%
- 8 Claude Sonnet 5 60.8%
- 9 GPT-5.6 Luna ↓2 60.4%
- 10
DeepSeek V4.1 Flash ↓5 56.6%
Report Comprehension human pass@1
6 models we track have a score here.
- 1 GPT-5.5 72%
- 2 Claude Sonnet 4.6 ↑4 62.5%
- 3 Claude Opus 4.7 ↑1 61.4%
- 4 GPT-5 mini ↑7 60.7%
- 5 GPT-5.1 ↑5 60.4%
- 6 GPT-5.4 Mini ↑3 42.2%
Report Comprehension human pass@5
6 models we track have a score here.
- 1 GPT-5.5 73.2%
- 2 Claude Sonnet 4.6 ↑4 71.7%
- 3 GPT-5 mini ↑8 66.6%
- 4 GPT-5.1 ↑6 64.9%
- 5 Claude Opus 4.7 ↓1 62.9%
- 6 GPT-5.4 Mini ↑3 48.7%
Static Timing Analysis pass@1
11 models we track have a score here.
- 1 GPT-5.6 Sol ↑1 72.7%
- 2
DeepSeek V4.1 Flash ↑3 72.7%
- 3 Claude Opus 4.7 ↑1 72.7%
- 4 GPT-5.5 ↓3 72.7%
- 5 Claude Opus 5 ↓2 63.6%
- 6 Claude Sonnet 4.6 63.6%
- 7 GPT-5.4 Mini ↑2 63.6%
- 8 Claude Sonnet 5 54.5%
- 9 GPT-5 mini ↑2 54.5%
- 10 GPT-5.1 54.5%
Static Timing Analysis pass@5
11 models we track have a score here.
- 1 Claude Sonnet 5 ↑7 72.7%
- 2 Claude Opus 5 ↑1 72.7%
- 3
DeepSeek V4.1 Flash ↑2 72.7%
- 4 Claude Opus 4.7 72.7%
- 5 GPT-5.5 ↓4 72.7%
- 6 GPT-5.6 Sol ↓4 63.6%
- 7 GPT-5 mini ↑4 63.6%
- 8 Claude Sonnet 4.6 ↓2 63.6%
- 9 GPT-5.4 Mini 63.6%
- 10 GPT-5.6 Luna ↓3 54.5%
Formal Verification pass@1
12 models we track have a score here.
- 1 GPT-5.6 Sol ↑1 81.8%
- 2 Claude Opus 5 ↑1 81.8%
- 3 GPT-5.5 ↓2 81.8%
- 4
DeepSeek V4.1 Flash ↑1 77.3%
- 5 Claude Opus 4.7 ↓1 72.7%
- 6 Claude Sonnet 4.6 63.6%
- 7 GPT-5.4 Mini ↑2 59.1%
- 8 GPT-5.6 Luna ↓1 59.1%
- 9 Claude Sonnet 5 ↓1 54.5%
- 10 GPT-5.1 54.5%
Formal Verification pass@5
12 models we track have a score here.
- 1
DeepSeek V4.1 Flash ↑4 95.5%
- 2 Claude Opus 5 ↑1 90.9%
- 3 GPT-5.5 ↓2 86.4%
- 4 GPT-5.6 Sol ↓2 81.8%
- 5 GPT-5.6 Luna ↑2 77.3%
- 6 GPT-5.4 Mini ↑3 72.7%
- 7 Claude Opus 4.7 ↓3 72.7%
- 8 Claude Sonnet 4.6 ↓2 68.2%
- 9 Claude Sonnet 5 ↓1 63.6%
- 10 GPT-5.1 63.6%
ECO pass@1
12 models we track have a score here.
- 1 GPT-5.6 Luna ↑6 60%
- 2 Claude Sonnet 4.6 ↑4 50%
- 3 GPT-5.4 Mini ↑6 50%
- 4 GPT-5.5 ↓3 50%
- 5 Claude Opus 4.7 ↓1 40%
- 6 Claude Sonnet 5 ↑2 40%
- 7 GPT-5.6 Sol ↓5 40%
- 8 Claude Opus 5 ↓5 40%
- 9
DeepSeek V4.1 Flash ↓4 40%
- 10 GPT-5 mini ↑1 30%
ECO pass@5
12 models we track have a score here.
- 1 GPT-5.5 60%
- 2 Claude Sonnet 5 ↑6 60%
- 3 Claude Sonnet 4.6 ↑3 50%
- 4 GPT-5.4 Mini ↑5 50%
- 5 Claude Opus 4.7 ↓1 50%
- 6 GPT-5.6 Luna ↑1 50%
- 7
DeepSeek V4.1 Flash ↓2 50%
- 8 GPT-5.6 Sol ↓6 40%
- 9 Claude Opus 5 ↓6 40%
- 10 GPT-5 mini ↑1 30%
P&R Vendor 1 pass@1
12 models we track have a score here.
- 1 GPT-5.6 Sol ↑1 52.3%
- 2 GPT-5.5 ↓1 45.5%
- 3 Claude Opus 5 39.8%
- 4 GPT-5.6 Luna ↑3 38.6%
- 5
DeepSeek V4.1 Flash 36.4%
- 6 Claude Sonnet 4.6 34.1%
- 7 Claude Opus 4.7 ↓3 34.1%
- 8 GPT-5.1 ↑2 31.8%
- 9 GPT-5.4 Mini 29.5%
- 10 Claude Sonnet 5 ↓2 27.3%
P&R Vendor 1 pass@5
12 models we track have a score here.
- 1 GPT-5.6 Sol ↑1 60.2%
- 2 GPT-5.5 ↓1 58%
- 3 Claude Opus 5 53.4%
- 4
DeepSeek V4.1 Flash ↑1 46.6%
- 5 GPT-5.6 Luna ↑2 45.5%
- 6 GPT-5.1 ↑4 43.2%
- 7 GPT-5.4 Mini ↑2 39.8%
- 8 Claude Opus 4.7 ↓4 37.5%
- 9 Claude Sonnet 4.6 ↓3 35.2%
- 10 Claude Sonnet 5 ↓2 34.1%
P&R Vendor 2 pass@1
12 models we track have a score here.
- 1 GPT-5.5 42.2%
- 2 GPT-5.6 Sol 41.1%
- 3 Claude Opus 5 40%
- 4 Claude Opus 4.7 36.7%
- 5 GPT-5.6 Luna ↑2 36.7%
- 6 Claude Sonnet 5 ↑2 33.3%
- 7
DeepSeek V4.1 Flash ↓2 32.2%
- 8 Claude Sonnet 4.6 ↓2 28.9%
- 9 GPT-5.1 ↑1 27.8%
- 10 GPT-5.4 Mini ↓1 26.7%
P&R Vendor 2 pass@5
12 models we track have a score here.
- 1 GPT-5.6 Sol ↑1 51.1%
- 2 GPT-5.5 ↓1 51.1%
- 3 GPT-5.6 Luna ↑4 46.7%
- 4 Claude Opus 5 ↓1 44.4%
- 5
DeepSeek V4.1 Flash 44.4%
- 6 Claude Opus 4.7 ↓2 41.1%
- 7 Claude Sonnet 5 ↑1 36.7%
- 8 GPT-5.4 Mini ↑1 35.6%
- 9 GPT-5.1 ↑1 31.1%
- 10 Claude Sonnet 4.6 ↓4 31.1%
Overall pass@5
12 models we track have a score here.
- 1 GPT-5.5 70.5%
- 2 Claude Opus 5 ↑1 67.4%
- 3
DeepSeek V4.1 Flash ↑2 64.8%
- 4 GPT-5.6 Sol ↓2 64.6%
- 5 Claude Opus 4.7 ↓1 62.6%
- 6 GPT-5.6 Luna ↑1 59.5%
- 7 Claude Sonnet 4.6 ↓1 59.4%
- 8 Claude Sonnet 5 58.6%
- 9 GPT-5.4 Mini 55.7%
- 10 GPT-5.1 54.3%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official PDAGENT-BENCH 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 PDAGENT-BENCH 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
- 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
All 10 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access Public tasks, plus a private test set.
- Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
- Pass Independent runs Scores from the official leaderboard (16 entries).
- Pass Models scored 12 models we track, plus 4 other entries on its leaderboard.
- Pass Newest score The newest is from September 2026, for DeepSeek V4.1 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 65.1%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on fenggev56.github.io: 12 rows for models we track and 4 other entries 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.