PhysDesignBench
GitHub
Reimplementation task · v2-canonical-j-diagnostic-gate

α-Si/Si₃N₄ Grating-v0

α-Si/Si3N4 SWG Grating Coupler Implementation/Optimization

Implement and optimize the apodized α-Si/Si₃N₄ SWG grating at 1310 nm. Trusted FDTD measures coupling loss and both bandwidths against the paper-derived reference envelope.

Runs
11
Scored finals
11 of 11
Best final
S = 0.27 Fable 5.1 · Claude Code
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: Fable 5.1 (Claude Code) 0.272; Sonnet 5.5 (high) (Claude Code) 0.266; GPT-6 Astra (Codex) 0.266; GPT-6.1 Sol (Codex) 0.262; Opus 5.5 (Claude Code) 0.261; Muse Wonder (high) (Muse Code) 0.228; GLM 5.3 Flash (max) (mini-SWE-agent) 0.219; Gemini 4 Argon (mini-SWE-agent) 0.217; GPT-5.6 Sol (Codex) 0.216; GLM 5.3 Flash (high) (Claude Code) 0.212; Kimi K3.1 Pro (Claude Code) 0.211.0.000.250.500.751.000153045607590105120135Elapsed time (min)Score SKimi K3.1 Pro · Claude Code: final S = 0.211 at 121 minGLM 5.3 Flash (high) · Claude Code: final S = 0.212 at 105 minGPT-5.6 Sol · Codex: final S = 0.216 at 120 minGemini 4 Argon · mini-SWE-agent: final S = 0.217 at 121 minGLM 5.3 Flash (max) · mini-SWE-agent: final S = 0.219 at 69 minMuse Wonder (high) · Muse Code: final S = 0.228 at 117 minOpus 5.5 · Claude Code: final S = 0.261 at 114 minGPT-6.1 Sol · Codex: final S = 0.262 at 120 minGPT-6 Astra · Codex: final S = 0.266 at 120 minSonnet 5.5 (high) · Claude Code: final S = 0.266 at 56 minFable 5.1 · Claude Code: final S = 0.272 at 73 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.

  • Optical mask

  • α-Si/Si₃N₄ Grating-v0 · Claude Code · Claude Fable 5.1

    S = 0.272 · scale bar 5 µm

    Fit
    Fable 5.1Claude CodeS = 0.27

  • α-Si/Si₃N₄ Grating-v0 · Claude Code · Claude Sonnet 5.5

    S = 0.266 · scale bar 5 µm

    Fit
    Sonnet 5.5 (high)Claude CodeS = 0.27

  • α-Si/Si₃N₄ Grating-v0 · Codex · GPT-6 Astra

    S = 0.266 · scale bar 5 µm

    Fit
    GPT-6 AstraCodexS = 0.27

  • α-Si/Si₃N₄ Grating-v0 · Codex · GPT-6.1-sol

    S = 0.262 · scale bar 5 µm

    Fit
    GPT-6.1 SolCodexS = 0.26

  • α-Si/Si₃N₄ Grating-v0 · Claude Code · Claude Opus 5.5

    S = 0.261 · scale bar 5 µm

    Fit
    Opus 5.5Claude CodeS = 0.26

  • α-Si/Si₃N₄ Grating-v0 · Muse Code · Muse Wonder

    S = 0.228 · scale bar 5 µm

    Fit
    Muse Wonder (high)Muse CodeS = 0.23

  • α-Si/Si₃N₄ Grating-v0 · mini-SWE-agent · GLM 5.3 Flash

    S = 0.219 · scale bar 5 µm

    Fit
    GLM 5.3 Flash (max)mini-SWE-agentS = 0.22

  • α-Si/Si₃N₄ Grating-v0 · mini-SWE-agent · Gemini-4-Argon

    S = 0.217 · scale bar 5 µm

    Fit
    Gemini 4 Argonmini-SWE-agentS = 0.22

  • α-Si/Si₃N₄ Grating-v0 · Codex · GPT-5.6-sol

    S = 0.216 · scale bar 5 µm

    Fit
    GPT-5.6 SolCodexS = 0.22

  • α-Si/Si₃N₄ Grating-v0 · Claude Code · GLM 5.3 Flash

    S = 0.212 · scale bar 5 µm

    Fit
    GLM 5.3 Flash (high)Claude CodeS = 0.21

  • α-Si/Si₃N₄ Grating-v0 · Claude Code · Kimi K3.1 Pro

    S = 0.211 · scale bar 5 µm

    Fit
    Kimi K3.1 ProClaude CodeS = 0.21