PhysDesignBench
GitHub
Reimplementation task · v5-independent-state-subruns

Sb₂Se₃/Si Rotator-v0

Sb2Se3/Si Reconfigurable Polarization Rotator Implementation/Optimization

Implement and optimize a reconfigurable Sb₂Se₃/Si polarization rotator. The trusted evaluator scores worst-case polarization-conversion loss over both material states and 1290–1330 nm.

Runs
11
Scored finals
11 of 11
Best final
S = 0.93 GPT-6 Astra · Codex
Score over timeBest-so-far score S for each run · small dots are probes, large dots final submissions
Score S from 0 to 1 against elapsed minutes. Final scores: GPT-6 Astra (Codex) 0.933; Opus 5.5 (Claude Code) 0.733; GPT-6.1 Sol (Codex) 0.649; Muse Wonder (high) (Muse Code) 0.519; Kimi K3.1 Pro (Claude Code) 0.499; Fable 5.1 (Claude Code) 0.497; GPT-5.6 Sol (Codex) 0.495; Sonnet 5.5 (high) (Claude Code) 0.466; Gemini 4 Argon (mini-SWE-agent) 0.265; GLM 5.3 Flash (high) (Claude Code) 0.265; GLM 5.3 Flash (max) (mini-SWE-agent) 0.145.0.000.250.500.751.000153045607590105120135Elapsed time (min)Score SGLM 5.3 Flash (max) · mini-SWE-agent: final S = 0.145 at 103 minGLM 5.3 Flash (high) · Claude Code: final S = 0.265 at 121 minGemini 4 Argon · mini-SWE-agent: final S = 0.265 at 121 minSonnet 5.5 (high) · Claude Code: final S = 0.466 at 31 minGPT-5.6 Sol · Codex: final S = 0.495 at 121 minFable 5.1 · Claude Code: final S = 0.497 at 78 minKimi K3.1 Pro · Claude Code: final S = 0.499 at 121 minMuse Wonder (high) · Muse Code: final S = 0.519 at 69 minGPT-6.1 Sol · Codex: final S = 0.649 at 120 minOpus 5.5 · Claude Code: final S = 0.733 at 68 minGPT-6 Astra · Codex: final S = 0.933 at 120 min

Showing 11 of 11 runs. Hover a line or a model to isolate it; click a model to hide or show it.

Final layouts

Every run’s final GDS, at one scale.

The design each run ended with, drawn from its evaluated GDS, with the task’s ports, modes and band as the evaluator defines them, the run’s own length, and the published result that S = 0.5 stands for. Every layout uses the same physical scale, so sizes compare directly. Hatched layouts failed a validity gate and score zero. Select a layout to zoom and toggle its layers.

  • Silicon core
  • Side Sb₂Se₃
  • Top Sb₂Se₃

  • Sb₂Se₃/Si Rotator-v0 · Codex · GPT-6 Astra

    S = 0.933 · scale bar 5 µm

    Fit
    GPT-6 AstraCodexS = 0.93

  • Sb₂Se₃/Si Rotator-v0 · Claude Code · Claude Opus 5.5

    S = 0.733 · scale bar 5 µm

    Fit
    Opus 5.5Claude CodeS = 0.73

  • Sb₂Se₃/Si Rotator-v0 · Codex · GPT-6.1-sol

    S = 0.649 · scale bar 5 µm

    Fit
    GPT-6.1 SolCodexS = 0.65

  • Sb₂Se₃/Si Rotator-v0 · Muse Code · Muse Wonder

    S = 0.519 · scale bar 5 µm

    Fit
    Muse Wonder (high)Muse CodeS = 0.52

  • Sb₂Se₃/Si Rotator-v0 · Claude Code · Kimi K3.1 Pro

    S = 0.499 · scale bar 5 µm

    Fit
    Kimi K3.1 ProClaude CodeS = 0.50

  • Sb₂Se₃/Si Rotator-v0 · Claude Code · Claude Fable 5.1

    S = 0.497 · scale bar 5 µm

    Fit
    Fable 5.1Claude CodeS = 0.50

  • Sb₂Se₃/Si Rotator-v0 · Codex · GPT-5.6-sol

    S = 0.495 · scale bar 5 µm

    Fit
    GPT-5.6 SolCodexS = 0.50

  • Sb₂Se₃/Si Rotator-v0 · Claude Code · Claude Sonnet 5.5

    S = 0.466 · scale bar 5 µm

    Fit
    Sonnet 5.5 (high)Claude CodeS = 0.47

  • Sb₂Se₃/Si Rotator-v0 · mini-SWE-agent · Gemini-4-Argon

    S = 0.265 · scale bar 5 µm

    Fit
    Gemini 4 Argonmini-SWE-agentS = 0.26

  • Sb₂Se₃/Si Rotator-v0 · Claude Code · GLM 5.3 Flash

    S = 0.265 · scale bar 5 µm

    Fit
    GLM 5.3 Flash (high)Claude CodeS = 0.26

  • Sb₂Se₃/Si Rotator-v0 · mini-SWE-agent · GLM 5.3 Flash

    S = 0.145 · scale bar 5 µm

    Fit
    GLM 5.3 Flash (max)mini-SWE-agentS = 0.15