AnalogPortBench
Models adapt analog circuit netlists to a different chip process or flag infeasible specifications. A simulator checks the designs against electrical limits.
The authors compare six models at two reasoning settings on 50 feasible tasks and four infeasible tasks. They report simulation feedback and independent attempts separately.
- Made by
- University of Delaware
- Released
- October 2026
- Kind
- Electronics
- Level
- Expert
- Scored
- Checked automatically
- Answers
- Code
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Multi-turn
- Refresh
- One fixed set
- Run by
- Anyone
- Strongest model tested
- Claude Opus 5.5 #5 on the capability index
- Last model added
- October 2026
- Scores refreshed
- November 20th, 2026
GPT-6.1 Sol leads with 56%
Best score per model 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.
Source: Official AnalogPortBench leaderboard
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-6.1 Sol | 56% | medium, Simulation feedback | Sep 2026 | Official leaderboard |
| 2 | Claude Opus 5.5 | 52% | medium, Simulation feedback | Sep 2026 | Official leaderboard |
| 3 | GPT-6 Sol | 34% | medium, Simulation feedback | Sep 2026 | Official leaderboard |
| 4 | Claude Sonnet 5.5 | 26% | medium, Simulation feedback | Sep 2026 | Official leaderboard |
| 5 | Claude Haiku 5.5 | 18% | medium, Simulation feedback | Oct 2026 | Official leaderboard |
| 6 | GPT-6 Luna | 12% | medium, Resampling | Sep 2026 | Official leaderboard |
Rankings by part
We read 9 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as T1 G1, T1 G2 and T1 G3. Claude Opus 5.5 leads 7 of them, and GPT-6.1 Sol leads the other 2. Arrows show how far a model moved from its overall rank among the models we track.
T1 G1
6 models we track have a score here.
- 1 Claude Opus 5.5 ↑1 93%
- 2 GPT-6 Sol ↑1 73%
- 3 GPT-6.1 Sol ↓2 73%
- 4 Claude Sonnet 5.5 60%
- 5 GPT-6 Luna ↑1 40%
- 6 Claude Haiku 5.5 ↓1 40%
T1 G2
6 models we track have a score here.
- 1 Claude Opus 5.5 ↑1 22%
- 2 GPT-6 Sol ↑1 11%
- 3 GPT-6.1 Sol ↓2 11%
- 4 GPT-6 Luna ↑2 0%
- 5 Claude Sonnet 5.5 ↓1 0%
- 6 Claude Haiku 5.5 ↓1 0%
T1 G3
6 models we track have a score here.
- 1 GPT-6.1 Sol 17%
- 2 Claude Opus 5.5 0%
- 3 GPT-6 Luna ↑3 0%
- 4 GPT-6 Sol ↓1 0%
- 5 Claude Sonnet 5.5 ↓1 0%
- 6 Claude Haiku 5.5 ↓1 0%
T2 G1
6 models we track have a score here.
- 1 Claude Opus 5.5 ↑1 73%
- 2 GPT-6 Sol ↑1 53%
- 3 GPT-6.1 Sol ↓2 53%
- 4 GPT-6 Luna ↑2 47%
- 5 Claude Sonnet 5.5 ↓1 40%
- 6 Claude Haiku 5.5 ↓1 33%
T2 G2
6 models we track have a score here.
- 1 Claude Opus 5.5 ↑1 0%
- 2 GPT-6 Luna ↑4 0%
- 3 GPT-6 Sol 0%
- 4 Claude Sonnet 5.5 0%
- 5 GPT-6.1 Sol ↓4 0%
- 6 Claude Haiku 5.5 ↓1 0%
T2 G3
6 models we track have a score here.
- 1 Claude Opus 5.5 ↑1 8.00%
- 2 GPT-6.1 Sol ↓1 8.00%
- 3 GPT-6 Luna ↑3 0%
- 4 GPT-6 Sol ↓1 0%
- 5 Claude Sonnet 5.5 ↓1 0%
- 6 Claude Haiku 5.5 ↓1 0%
T3 G1
6 models we track have a score here.
- 1 GPT-6.1 Sol 21%
- 2 Claude Opus 5.5 15%
- 3 GPT-6 Luna ↑3 0%
- 4 GPT-6 Sol ↓1 0%
- 5 Claude Sonnet 5.5 ↓1 0%
- 6 Claude Haiku 5.5 ↓1 0%
T3 G2
6 models we track have a score here.
- 1 Claude Opus 5.5 ↑1 0%
- 2 GPT-6 Luna ↑4 0%
- 3 GPT-6 Sol 0%
- 4 Claude Sonnet 5.5 0%
- 5 GPT-6.1 Sol ↓4 0%
- 6 Claude Haiku 5.5 ↓1 0%
T3 G3
6 models we track have a score here.
- 1 Claude Opus 5.5 ↑1 0%
- 2 GPT-6 Luna ↑4 0%
- 3 GPT-6 Sol 0%
- 4 Claude Sonnet 5.5 0%
- 5 GPT-6.1 Sol ↓4 0%
- 6 Claude Haiku 5.5 ↓1 0%
More effort adds a median 23 points
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.
Source: Official AnalogPortBench leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-6.1 Sol | medium, Simulation feedback | 56% | |||
| Claude Opus 5.5 | medium, Simulation feedback | 52% | |||
| GPT-6 Sol | medium, Simulation feedback | 34% | |||
| Claude Opus 5.5 | medium, Resampling | 32% | |||
| GPT-6.1 Sol | medium, Resampling | 30% | |||
| Claude Sonnet 5.5 | medium, Simulation feedback | 26% | |||
| Claude Opus 5.5 | medium, zero-shot | 22% | |||
| GPT-6 Sol | medium, Resampling | 20% | |||
| GPT-6.1 Sol | medium, zero-shot | 18.7% | |||
| Claude Haiku 5.5 | medium, Simulation feedback | 18% | |||
| Claude Opus 5.5 | low, zero-shot | 14.7% | |||
| GPT-6 Sol | medium, zero-shot | 13.3% | |||
| GPT-6 Luna | medium, Resampling | 12% | |||
| GPT-6 Luna | medium, Simulation feedback | 12% | |||
| Claude Sonnet 5.5 | medium, Resampling | 12% | |||
| GPT-6.1 Sol | low, zero-shot | 12% | |||
| GPT-6 Sol | low, zero-shot | 10.7% | |||
| Claude Sonnet 5.5 | low, zero-shot | 10% | |||
| Claude Sonnet 5.5 | medium, zero-shot | 10% | |||
| GPT-6 Luna | medium, zero-shot | 8.67% | |||
| GPT-6 Luna | low, zero-shot | 8.00% | |||
| Claude Haiku 5.5 | medium, Resampling | 8.00% | |||
| Claude Haiku 5.5 | low, zero-shot | 6.67% | |||
| Claude Haiku 5.5 | medium, zero-shot | 6.00% |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official AnalogPortBench 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 AnalogPortBench 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 11 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 5 pass
- Pass Independent runs Scores from the official leaderboard (6 entries).
- Pass Models scored 6 models we track.
- Pass Newest score The newest is from October 2026, for Claude Haiku 5.5.
- Pass Frontier models 3 of the top 10 have a score, led by Claude Opus 5.5 (#5).
- Pass Headroom The best is 56%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on saeidrajabii.github.io: 24 rows for models we track 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.