Benchmark Almanac

HWE-Bench v1.0 (RTL bug repair)

GPT-5.6 Sol 79.9% leads since July 2026
Not in model cards
Version 1.0

Agents fix bugs in six open-source hardware projects. The benchmark reports the share of fixes that pass repository tests in simulation.

The original release contains 417 tasks. The authors retain the v1.0 dataset tag and scores for comparisons.

Released
April 2026
Kind
Electronics
Level
Expert
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Code → Code, Actions
Interaction
Agentic
Environment
Terminal, A simulator
Items from
Real work and use
Run by
Anyone
In views
Agents, Robotics
Strongest model tested
GPT-5.6 Sol #14 on the capability index
Last model added
July 2026
Scores refreshed
November 20th, 2026
15
Models scored on its own leaderboard

GPT-5.6 Sol leads with 79.9% among models released by July 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 July 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 GPT-5.6 Sol max 79.9%
  2. 2 GPT-5.5 xhigh 76.5%
  3. 3 DeepSeek V4 Flash 0731 max 76.5%
  4. 4 Claude Opus 4.7 max 74.6%
  5. 5 GLM-5.2 max 73.6%
  6. 6 GPT-5.4 xhigh 70.7%
  7. 7 Claude Opus 4.6 high 68.1%
  8. 8 Kimi K2.6 66.9%
  9. 9 Claude Sonnet 4.6 high 66.7%
  10. 10 GLM-5.1 63.1%
  11. 11 DeepSeek-V4-Pro max 60.4%
  12. 12 DeepSeek-V4-Flash max 58.3%
  13. 13 Qwen 3.6 Plus 52.5%
  14. 14 DeepSeek-V3.2 52%
  15. 15 Kimi K2.5 47.7%

Source: Official HWE-Bench v1.0 (RTL bug repair) leaderboard

#ModelScoreSettingDateSource
1 GPT-5.6 Sol 79.9% max, Codex CLI Jul 2026 Official leaderboard
2 GPT-5.5 76.5% xhigh, Codex CLI Apr 2026 Official leaderboard
3 DeepSeek V4 Flash 0731 76.5% max, OpenHands SDK Jul 2026 Official leaderboard
4 Claude Opus 4.7 74.6% max, Claude Code Apr 2026 Official leaderboard
5 GLM-5.2 73.6% max, OpenHands SDK Jun 2026 Official leaderboard
6 GPT-5.4 70.7% xhigh, Codex CLI Mar 2026 Official leaderboard
7 Claude Opus 4.6 68.1% high, Claude Code Feb 2026 Official leaderboard
8 Kimi K2.6 66.9% Kimi CLI Apr 2026 Official leaderboard
9 Claude Sonnet 4.6 66.7% high, Claude Code Feb 2026 Official leaderboard
10 GLM-5.1 63.1% OpenHands SDK Apr 2026 Official leaderboard
11 DeepSeek-V4-Pro 60.4% max, Claude Code Apr 2026 Official leaderboard
12 DeepSeek-V4-Flash 58.3% max, Claude Code Apr 2026 Official leaderboard
13 Qwen 3.6 Plus 52.5% OpenHands SDK Mar 2026 Official leaderboard
14 DeepSeek-V3.2 52% OpenHands SDK Dec 2025 Official leaderboard
15 Kimi K2.5 47.7% Kimi CLI Jan 2026 Official leaderboard

Rankings by part

We read 6 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as OpenTitan, Ibex and CVA6. GPT-5.6 Sol leads 2 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

OpenTitan

15 models we track have a score here.

  1. 1 GPT-5.6 Sol 75.1%
  2. 2 DeepSeek V4 Flash 0731 ↑1 71.8%
  3. 3 GPT-5.5 ↓1 71%
  4. 4 Claude Opus 4.7 70.6%
  5. 5 GLM-5.2 70.2%
  6. 6 GPT-5.4 64.1%
  7. 7 Claude Opus 4.6 63.7%
  8. 8 Claude Sonnet 4.6 ↑1 62%
  9. 9 GLM-5.1 ↑1 62%
  10. 10 Kimi K2.6 ↓2 61.6%

The best score, 79.9%, costs $1,179 for the full run

Runs on the cost frontier score higher than any run that's cheaper. DeepSeek V4 Flash 0731 at max comes within 3.4 points of the best for $40 for the full run, 30 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

DeepSeek OpenAI Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $10 $100 $1,000 $10,000 Cost of the full run, log scale GPT-5.6 Sol, max DeepSeek V4 Flash 0731, max DeepSeek-V4-Pro, max 0% 20% 40% 60% 80% 100% $10 $100 $1,000 $10,000 Cost of the full run, log scale GPT-5.6 Sol, max DeepSeek-V4-Pro, max

Source: Official HWE-Bench v1.0 (RTL bug repair) leaderboard

ModelEffortHarnessScoreCost of the full runOutput tokens
GPT-5.6 Sol max Codex CLI 79.9% $1,179
GPT-5.5 xhigh Codex CLI 76.5% $1,064
DeepSeek V4 Flash 0731 max OpenHands SDK 76.5% $40
Claude Opus 4.7 max Claude Code 74.6% $561
GLM-5.2 max OpenHands SDK 73.6% $998
GPT-5.4 xhigh Codex CLI 70.7% $755
Claude Opus 4.6 high Claude Code 68.1% $213
Kimi K2.6 Kimi CLI 66.9% $240
Claude Sonnet 4.6 high Claude Code 66.7% $108
GLM-5.1 OpenHands SDK 63.1% $225
DeepSeek-V4-Pro max Claude Code 60.4% $33
DeepSeek-V4-Flash max Claude Code 58.3% $12
Qwen 3.6 Plus OpenHands SDK 52.5% $209
DeepSeek-V3.2 OpenHands SDK 52% $42
Kimi K2.5 Kimi CLI 47.7% $67

4 models have held number one on the official leaderboard since its release in April 2026

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

OpenAI Anthropic
Own leaderboard Replaced by v1.1 (RTL bug repair), Nov 2026
GPT-5.6 Sol 79.9%

Source: Official HWE-Bench v1.0 (RTL bug repair) leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-5.6 Sol GPT-5.5 GPT-5.4 DeepSeek-V3.2 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-5.6 Sol GPT-5.4

Source: Official HWE-Bench v1.0 (RTL bug repair) 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.

Versions

We switch the top row to each new version when it comes out. Open it for one row per version. An old version stays in play while people still test it, so whoever first takes it past 85% gets credit for closing it out.

OpenAI Anthropic
HWE-Bench v1.1 (RTL bug repair) Passed 85% in Nov 2026, by Claude Opus 5.5
HWE-Bench v1.0 (RTL bug repair) Replaced by v1.1 (RTL bug repair), Nov 2026
GPT-5.6 Sol 79.9%
HWE-Bench v1.1 (RTL bug repair) Passed 85% in Nov 2026, by Claude Opus 5.5

Source: Official leaderboards

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 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (15 entries).
  • Pass Models scored 15 models we track.
  • Pass Newest score The newest is from July 2026, for DeepSeek V4 Flash 0731.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is GPT-5.6 Sol (#14).
  • Pass Headroom The best is 79.9%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on pku-liang.github.io: 15 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.

How we check