Benchmark Almanac

VerilogEval v2

Gemma 4 31B IT 88% leads since April 2026
Not in model cards

Models write Verilog hardware descriptions from specifications. The hub port uses the specification-to-RTL tasks, where simulation checks generated designs against reference behavior.

In February 2025, GPT-4o led specification-to-RTL tasks at 63% pass rate. Llama 3.1 405B reached 58%, and RTL-Coder 6.7B reached 34%.

Made by
NVIDIA and collaborators
Released
August 2024
Kind
Electronics
Level
University
Scored
Checked automatically
Answers
Code
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Code → Code
Interaction
Single turn
Run by
Anyone
Based on
VerilogEval
Strongest model tested
GLM-5.2 #55 on the capability index
Last model added
June 2026
Scores refreshed
November 20th, 2026
44
Entries on TuRTLe’s leaderboard, 19 models we track

Gemma 4 31B IT leads with 88% among models released by June 2026 Gemma 4 31B IT leads with 88% among models released by June 2026

Best score per entry from TuRTLe’s 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 June 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Gemma 4 31B IT 88%
  2. 2 Kimi K2.5 87.2%
  3. 3 GLM-5-FP8 85.4%
  4. 4 GLM-5.2 85%
  5. 5 DeepSeek-R1 May 2025 81.5%
  6. 6 gpt-oss-120b 78.1%
  7. 7 Qwen3 236B A22B 76.9%
  8. 8 DeepSeek-V3.1-Terminus 76.8%
  9. 9 Kimi K2 Sep 2025 72.8%
  10. 10 Seed-OSS-36B 71.2%
  11. 11 Gemini 2.5 Flash Sep 2025, medium 70.4%
  12. 12 gpt-oss-20b 69.7%
  13. 13 QwQ-32B 66.4%
  14. 14 Qwen3-Coder-480B-A35B 65.4%
  15. 15 InCoder-32B-Thinking 64.4%
  16. 16 InCoder-32B 64.1%
  17. 17 Llama 3.1-405B 59%
  18. 18 Hermes-4-14B-Reasoning 55.9%
  19. 19 OriGen 54.2%
  20. 20 Seed-Coder-8B-Instruct 53.5%
  21. 21 Qwen2.5-32B 52.6%
  22. 22 Qwen2.5-72B 52.4%
  23. 23 Qwen3-8B 50%
  24. 24 SeedCoder 8B Reasoning 49.2%
  25. 25 HaVen-CodeQwen 47.3%
  26. 26 QwenCoder 2.5 32B 45%
  27. 27 Magistral Small 2506 43%
  28. 28 Hermes-4-14B 42.3%
  29. 29 Llama 3.(1-3) 70B 40.6%
  30. 30 RTLCoder DeepSeek 39.5%
  31. 31 StarChat2 15B v0.1 38%
  32. 32 QwenCoder 2.5 14B 37.8%
  33. 33 CodeV R1 Distill Qwen 7B 33.3%
  34. 34 CodeLlama 70B 33.1%
  35. 35 DeepSeek Coder 6.7B 29.6%
  36. 36 OpenCoder 8B 28.5%
  37. 37 DeepCoder 14B 28.1%
  38. 38 DeepSeek-R1-Distill-Qwen-14B 25.5%
  39. 39 RTLCoder Mistral 24.6%
  40. 40 DeepSeek Coder 33B 23.3%
  41. 41 CodeV-QW-7B 19.6%
  42. 42 CodeV-DS-6.7B 15%
  43. 43 CodeV-CL-7B 13.1%
  44. 44 QwenCoder 2.5 7B 6.92%

Source: TuRTLe’s leaderboard

#EntryScoreSettingDateSource
1 Gemma 4 31B IT 88% Apr 2026 Official leaderboard
2 Kimi K2.5 87.2% Jan 2026 Official leaderboard
3 GLM-5-FP8 85.4% Official leaderboard
4 GLM-5.2 85% Jun 2026 Official leaderboard
5 DeepSeek-R1 May 2025 81.5% May 2025 Official leaderboard
6 gpt-oss-120b 78.1% Aug 2025 Official leaderboard
7 Qwen3 236B A22B 76.9% Official leaderboard
8 DeepSeek-V3.1-Terminus 76.8% Sep 2025 Official leaderboard
9 Kimi K2 Sep 2025 72.8% Sep 2025 Official leaderboard
10 Seed-OSS-36B 71.2% Official leaderboard
11 Gemini 2.5 Flash Sep 2025 70.4% medium Sep 2025 Official leaderboard
12 gpt-oss-20b 69.7% Aug 2025 Official leaderboard
13 QwQ-32B 66.4% Mar 2025 Official leaderboard
14 Qwen3-Coder-480B-A35B 65.4% Jul 2025 Official leaderboard
15 InCoder-32B-Thinking 64.4% Official leaderboard
16 InCoder-32B 64.1% Official leaderboard
17 Llama 3.1-405B 59% Jul 2024 Official leaderboard
18 Hermes-4-14B-Reasoning 55.9% Official leaderboard
19 OriGen 54.2% Official leaderboard
20 Seed-Coder-8B-Instruct 53.5% May 2025 Official leaderboard
21 Qwen2.5-32B 52.6% Sep 2024 Official leaderboard
22 Qwen2.5-72B 52.4% Sep 2024 Official leaderboard
23 Qwen3-8B 50% Apr 2025 Official leaderboard
24 SeedCoder 8B Reasoning 49.2% Official leaderboard
25 HaVen-CodeQwen 47.3% Official leaderboard
26 QwenCoder 2.5 32B 45% Official leaderboard
27 Magistral Small 2506 43% Official leaderboard
28 Hermes-4-14B 42.3% Official leaderboard
29 Llama 3.(1-3) 70B 40.6% Official leaderboard
30 RTLCoder DeepSeek 39.5% Official leaderboard
31 StarChat2 15B v0.1 38% Official leaderboard
32 QwenCoder 2.5 14B 37.8% Official leaderboard
33 CodeV R1 Distill Qwen 7B 33.3% Official leaderboard
34 CodeLlama 70B 33.1% Official leaderboard
35 DeepSeek Coder 6.7B 29.6% Nov 2023 Official leaderboard
36 OpenCoder 8B 28.5% Official leaderboard
37 DeepCoder 14B 28.1% Official leaderboard
38 DeepSeek-R1-Distill-Qwen-14B 25.5% Jan 2025 Official leaderboard
39 RTLCoder Mistral 24.6% Official leaderboard
40 DeepSeek Coder 33B 23.3% Nov 2023 Official leaderboard
41 CodeV-QW-7B 19.6% Official leaderboard
42 CodeV-DS-6.7B 15% Official leaderboard
43 CodeV-CL-7B 13.1% Official leaderboard
44 QwenCoder 2.5 7B 6.92% Official leaderboard
#ModelScoreSettingDateSource
1 Gemma 4 31B IT 88% Apr 2026 Official leaderboard
2 Kimi K2.5 87.2% Jan 2026 Official leaderboard
3 GLM-5.2 85% Jun 2026 Official leaderboard
4 DeepSeek-R1 May 2025 81.5% May 2025 Official leaderboard
5 gpt-oss-120b 78.1% Aug 2025 Official leaderboard
6 DeepSeek-V3.1-Terminus 76.8% Sep 2025 Official leaderboard
7 Kimi K2 Sep 2025 72.8% Sep 2025 Official leaderboard
8 Gemini 2.5 Flash Sep 2025 70.4% medium Sep 2025 Official leaderboard
9 gpt-oss-20b 69.7% Aug 2025 Official leaderboard
10 QwQ-32B 66.4% Mar 2025 Official leaderboard
11 Qwen3-Coder-480B-A35B 65.4% Jul 2025 Official leaderboard
12 Llama 3.1-405B 59% Jul 2024 Official leaderboard
13 Seed-Coder-8B-Instruct 53.5% May 2025 Official leaderboard
14 Qwen2.5-32B 52.6% Sep 2024 Official leaderboard
15 Qwen2.5-72B 52.4% Sep 2024 Official leaderboard
16 Qwen3-8B 50% Apr 2025 Official leaderboard
17 DeepSeek Coder 6.7B 29.6% Nov 2023 Official leaderboard
18 DeepSeek-R1-Distill-Qwen-14B 25.5% Jan 2025 Official leaderboard
19 DeepSeek Coder 33B 23.3% Nov 2023 Official leaderboard

5 models have held number one on TuRTLe’s leaderboard since its release in August 2024

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

Meta Qwen DeepSeek Moonshot AI Google DeepMind
TuRTLe leaderboard Passed 85% in Jan 2026, by Kimi K2.5

Source: TuRTLe’s leaderboard

Kimi K2.5 on the official leaderboard passed 85% in January 2026

That was 17 months after its release. By the same age, 29% of the benchmarks from 2024 had passed 85%. All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On TuRTLe’s leaderboard Best so far
0% 20% 40% 60% 80% 100% 2024 2025 2026 Near the ceiling (85%) Gemma 4 31B IT DeepSeek-R1 QwQ-32B Llama 3.1-405B DeepSeek Coder 6.7B 0% 20% 40% 60% 80% 100% 2024 2025 2026 Near the ceiling (85%) Gemma 4 31B IT DeepSeek-R1

Source: TuRTLe’s 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 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (44 entries).
  • Pass Models scored 19 models we track, plus 25 other entries on its leaderboard.
  • Pass Newest score The newest is from June 2026, for GLM-5.2.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is GLM-5.2 (#55).
  • Note Headroom Saturated. The best is 88%, past 85% since January 2026.
Leaderboard import 3 pass
  • Pass Importer We read the board on TuRTLe: 20 rows for models we track and 25 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.

How we check