OmniRouting
Models route copper traces between pads on printed circuit boards. The evaluator counts required nets that connect without design rule violations.
Nine models have one-shot scores and nine have agent scores. The two tables share eight models. The site links sample data.
- Made by
- Pennsylvania State University and collaborators
- Released
- July 2026
- Kind
- Electronics
- Level
- University
- Scored
- Checked automatically
- Answers
- Agent task
- Measures
- Capability
- Language
- English
- Interaction
- Agentic
- Environment
- APIs and tools
- Items from
- Real work and use, Written by experts
- Refresh
- One fixed set
- Run by
- A host
- Strongest model tested
- GPT-5.5 #21 on the capability index
- Last model added
- May 2026
- Scores refreshed
- November 20th, 2026
GPT-5.5 leads with 28% among models released by May 2026
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.
The newest model with a score here came out in May 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
- GPT-6 Astra
- and 16 more
- 1 GPT-5.5 All Tools 28%
- 2 Claude Opus 4.8 All Tools 14.9%
- 3 GPT-5 mini Visualization 9.91%
- 4
Qwen3-235B-A22B Base 5.23%
- 5
Llama 4 Maverick Base 4.28%
- 6
Gemini 3.1 Pro Prerouting / Remake 4.08%
- 7
Mistral Large 3 Unlimited layer 3.68%
- 8
Qwen3.5-9B Visualization 1.85%
- 9
Gemini 2.5 Flash-Lite Sep 2025, Base 1.37%
- 10
Ministral 3 14B All Tools 0.96%
Source: Official OmniRouting leaderboard
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.5 | 28% | All Tools, Agent | Apr 2026 | Official leaderboard |
| 2 | Claude Opus 4.8 | 14.9% | All Tools, Agent | May 2026 | Official leaderboard |
| 3 | GPT-5 mini | 9.91% | Visualization, Agent | Aug 2025 | Official leaderboard |
| 4 | | 5.23% | Base, One-shot | Apr 2025 | Official leaderboard |
| 5 | | 4.28% | Base, One-shot | Apr 2025 | Official leaderboard |
| 6 | | 4.08% | Prerouting / Remake, Agent | Feb 2026 | Official leaderboard |
| 7 | | 3.68% | Unlimited layer, One-shot | Dec 2025 | Official leaderboard |
| 8 | | 1.85% | Visualization, Agent | Feb 2026 | Official leaderboard |
| 9 | | 1.37% | Base, One-shot | Sep 2025 | Official leaderboard |
| 10 | | 0.96% | All Tools, Agent | Dec 2025 | 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.
Source: Official OmniRouting leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-5.5 | All Tools | Agent | 28% | ||
| GPT-5.5 | Routing Score | Agent | 24.8% | ||
| GPT-5.5 | Prerouting / Remake | Agent | 22.6% | ||
| GPT-5.5 | Semantic Input | Agent | 21.8% | ||
| GPT-5.5 | No Tools | Agent | 21.7% | ||
| GPT-5.5 | Visualization | Agent | 21% | ||
| Claude Opus 4.8 | All Tools | Agent | 14.9% | ||
| Claude Opus 4.8 | Prerouting / Remake | Agent | 14.1% | ||
| Claude Opus 4.8 | Routing Score | Agent | 13.4% | ||
| Claude Opus 4.8 | Base | One-shot | 12.6% | ||
| Claude Opus 4.8 | Prerouter | One-shot | 12.5% | ||
| Claude Opus 4.8 | Netclass detail | One-shot | 12.4% | ||
| Claude Opus 4.8 | Unlimited layer | One-shot | 12.1% | ||
| Claude Opus 4.8 | Semantic Input | Agent | 11.3% | ||
| Claude Opus 4.8 | Visualization | Agent | 11.2% | ||
| Claude Opus 4.8 | No Tools | Agent | 10.7% | ||
| Claude Opus 4.8 | Fewshot | One-shot | 10.4% | ||
| GPT-5.5 | Prerouter | One-shot | 9.99% | ||
| GPT-5 mini | Visualization | Agent | 9.91% | ||
| GPT-5.5 | Netclass detail | One-shot | 9.87% | ||
| GPT-5 mini | No Tools | Agent | 9.86% | ||
| GPT-5 mini | Routing Score | Agent | 9.85% | ||
| GPT-5.5 | Fewshot | One-shot | 9.42% | ||
| GPT-5 mini | Prerouting / Remake | Agent | 9.15% | ||
| GPT-5.5 | Base | One-shot | 9.13% | ||
| GPT-5 mini | Semantic Input | Agent | 9.08% | ||
| GPT-5.5 | Unlimited layer | One-shot | 9.00% | ||
| GPT-5 mini | All Tools | Agent | 6.82% | ||
| GPT-5 mini | Prerouter | One-shot | 5.91% | ||
| GPT-5 mini | Base | One-shot | 5.75% | ||
| GPT-5 mini | Netclass detail | One-shot | 5.57% | ||
| Base | One-shot | 5.23% | |||
| GPT-5 mini | Unlimited layer | One-shot | 5.12% | ||
| Unlimited layer | One-shot | 4.40% | |||
| Base | One-shot | 4.28% | |||
| Netclass detail | One-shot | 4.16% | |||
| GPT-5 mini | Fewshot | One-shot | 4.09% | ||
| Prerouting / Remake | Agent | 4.08% | |||
| Unlimited layer | One-shot | 3.68% | |||
| Base | One-shot | 3.67% | |||
| Netclass detail | One-shot | 3.67% | |||
| Prerouter | One-shot | 3.64% | |||
| Prerouter | One-shot | 3.63% | |||
| Unlimited layer | One-shot | 3.58% | |||
| Netclass detail | One-shot | 3.37% | |||
| Fewshot | One-shot | 3.13% | |||
| Prerouter | One-shot | 3.03% | |||
| Fewshot | One-shot | 2.90% | |||
| Semantic Input | Agent | 2.59% | |||
| No Tools | Agent | 2.20% | |||
| Visualization | Agent | 1.85% | |||
| Visualization | Agent | 1.82% | |||
| Prerouting / Remake | Agent | 1.40% | |||
| No Tools | Agent | 1.40% | |||
| Routing Score | Agent | 1.39% | |||
| Base | One-shot | 1.37% | |||
| Netclass detail | One-shot | 1.33% | |||
| Unlimited layer | One-shot | 1.28% | |||
| Base | One-shot | 1.24% | |||
| All Tools | Agent | 1.17% | |||
| Unlimited layer | One-shot | 1.12% | |||
| All Tools | Agent | 1.11% | |||
| Netclass detail | One-shot | 1.10% | |||
| Fewshot | One-shot | 1.08% | |||
| Prerouter | One-shot | 1.07% | |||
| All Tools | Agent | 0.96% | |||
| Prerouting / Remake | Agent | 0.80% | |||
| Routing Score | Agent | 0.79% | |||
| Prerouter | One-shot | 0.77% | |||
| Base | One-shot | 0.71% | |||
| Visualization | Agent | 0.49% | |||
| Prerouting / Remake | Agent | 0.47% | |||
| All Tools | Agent | 0.46% | |||
| Fewshot | One-shot | 0.44% | |||
| Routing Score | Agent | 0.44% | |||
| No Tools | Agent | 0.39% | |||
| Netclass detail | One-shot | 0.37% | |||
| No Tools | Agent | 0.33% | |||
| Semantic Input | Agent | 0.27% | |||
| No Tools | Agent | 0.27% | |||
| Semantic Input | Agent | 0.27% | |||
| Visualization | Agent | 0.26% | |||
| Routing Score | Agent | 0.23% | |||
| Routing Score | Agent | 0.20% | |||
| Prerouting / Remake | Agent | 0.15% | |||
| All Tools | Agent | 0.12% | |||
| Visualization | Agent | 0.11% | |||
| Fewshot | One-shot | 0.09% | |||
| Semantic Input | Agent | 0.09% | |||
| Semantic Input | Agent | 0.09% | |||
| No Tools | Agent | 0% |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official OmniRouting 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 OmniRouting 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
8 of 10 checks pass, 2 warnings. 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 1 warning 3 pass 1 note
- Pass Independent runs Scores from the official leaderboard (10 entries).
- Pass Models scored 10 models we track.
- Warning Newest score The newest came 6 months ago, in May 2026, for Claude Opus 4.8.
- Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.5 (#21).
- Pass Headroom The best is 28%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on omnieda.com: 91 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.