Benchmark Almanac

AssertLLM2

GPT-5.2 95% leads since May 2026
Not in model cards

Models write assertions for hardware specifications. A formal verifier checks which assertions hold, which logic they cover, and which deliberately introduced faults they detect.

ChipOmni ran DeepSeek on all 83 designs. The maintainers also release a 25-design subset spanning all 13 categories.

Released
May 2026
Kind
Electronics
Level
Expert
Scored
Checked automatically
Answers
Code
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Code → Code
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Strongest model tested
GPT-5.2 #51 on the capability index
Last model added
December 2025
Scores refreshed
November 20th, 2026
6
Entries in its paper, 5 models we track

GPT-5.2 leads with 95% among models released by December 2025 GPT-5.2 leads with 95% among models released by December 2025

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. The paper reports no overall score, so this is the percentage of syntactically valid assertions, its first metric. 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. This view shows only models we track, each at its best.

The newest model with a score here came out in December 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-5.2 Average 95%
  2. 2 Claude Sonnet 4.5 Average 93%
  3. 3 Llama 4 Maverick Average 89.2%
  4. 4 DeepSeek-V3.2 Average 88.9%
  5. 5 Gemini-2.5-Pro (snapshot unspecified) Average 88.3%
  6. 6 Qwen3-Coder-Plus Average 77.9%

Source: The AssertLLM2 paper

#EntryScoreSettingDateSource
1 GPT-5.2 95% Average Dec 2025 Official leaderboard
2 Claude Sonnet 4.5 93% Average Sep 2025 Official leaderboard
3 Llama 4 Maverick 89.2% Average Apr 2025 Official leaderboard
4 DeepSeek-V3.2 88.9% Average Dec 2025 Official leaderboard
5 Gemini-2.5-Pro (snapshot unspecified) 88.3% Average Official leaderboard
6 Qwen3-Coder-Plus 77.9% Average Sep 2025 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-5.2 95% Average Dec 2025 Official leaderboard
2 Claude Sonnet 4.5 93% Average Sep 2025 Official leaderboard
3 Llama 4 Maverick 89.2% Average Apr 2025 Official leaderboard
4 DeepSeek-V3.2 88.9% Average Dec 2025 Official leaderboard
5 Qwen3-Coder-Plus 77.9% Average Sep 2025 Official leaderboard

Rankings by part

We read 7 parts from the benchmark’s paper, each ranking models on one part of the test, such as Syntax, Proven assertions and Syntactically correct assertions. Claude Sonnet 4.5 leads 5 of them, and GPT-5.2 leads the other 2. Arrows show how far a model moved from its overall rank among the models we track.

Syntax

5 models we track have a score here.

  1. 1 GPT-5.2 95%
  2. 2 Claude Sonnet 4.5 93%
  3. 3 Llama 4 Maverick 89.2%
  4. 4 DeepSeek-V3.2 88.9%
  5. 5 Qwen3-Coder-Plus 77.9%

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.

Lowest effort to highest, light to dark
GPT-5.2
Average
95%
Claude Sonnet 4.5
Average
93%
Llama 4 Maverick
Average
89.2%
DeepSeek-V3.2
Average
88.9%
Qwen3-Coder-Plus
Average
77.9%

Source: The AssertLLM2 paper

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.2 Average 95%
GPT-5.2 Union 95%
Claude Sonnet 4.5 Average 93%
Claude Sonnet 4.5 Union 93%
Llama 4 Maverick Average 89.2%
Llama 4 Maverick Union 89.2%
DeepSeek-V3.2 Average 88.9%
DeepSeek-V3.2 Union 88.9%
Qwen3-Coder-Plus Average 77.9%
Qwen3-Coder-Plus Union 77.9%

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.

Lowest effort to highest, light to dark
GPT-5.2
Average
95%
Claude Sonnet 4.5
Average
93%
Llama 4 Maverick
Average
89.2%
DeepSeek-V3.2
Average
88.9%
Qwen3-Coder-Plus
Average
77.9%

Source: The AssertLLM2 paper

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.2 Average 95%
GPT-5.2 Union 95%
Claude Sonnet 4.5 Average 93%
Claude Sonnet 4.5 Union 93%
Llama 4 Maverick Average 89.2%
Llama 4 Maverick Union 89.2%
DeepSeek-V3.2 Average 88.9%
DeepSeek-V3.2 Union 88.9%
Qwen3-Coder-Plus Average 77.9%
Qwen3-Coder-Plus Union 77.9%

Number one over time

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

OpenAI
Benchmark paper Passed 85% in May 2026, by GPT-5.2
GPT-5.2 95%

Source: The AssertLLM2 paper

Llama 4 Maverick on the official leaderboard passed 85% in April 2025

Models released before it came out already scored that high. All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) GPT-5.2 Claude Sonnet 4.5 Llama 4 Maverick 0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) GPT-5.2 Llama 4 Maverick

Source: Official AssertLLM2 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

All 112 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 8 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
  • Pass Independent runs Scores from the benchmark’s paper (6 entries).
  • Pass Models scored 5 models we track, plus 1 other entry on its leaderboard.
  • Note Newest score The newest came 11 months ago, in December 2025, for GPT-5.2. That’s expected for a saturated test.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is GPT-5.2 (#51).
  • Note Headroom Saturated. The best is 95%, past 85% since April 2025.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 10 rows for models we track and 1 other entry 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.

How we check