PhysDesignBench
GitHub
Design task · v1

MZI-Spec-8

Compact PDK-Physical Eight-Delay Sparse-Spectrum MZI Spectrometer

Design a compact eight-delay MZI spectrometer with physical SiEPIC couplers, a routed 1x8 splitter tree, 16 grating outputs, and trusted sparse-spectrum reconstruction scoring.

Runs
13
Scored finals
13 of 13
Best final
S = 0.51 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) 0.508; Gemini 4 Argon (mini-SWE-agent) 0.504; GPT-6.1 Sol (Codex) 0.498; GPT-6.1 Sol (mini-SWE-agent) 0.494; Fable 5.1 (Claude Code) 0.494; Sonnet 5.5 (mini-SWE-agent) 0.487; GLM 5.3 Flash (max) (mini-SWE-agent) 0.484; Opus 5.5 (Claude Code) 0.483; Muse Wonder (high) (Muse Code) 0.480; Kimi K3.1 Pro (Claude Code) 0.473; Sonnet 5.5 (high) (Claude Code) 0.465; GLM 5.3 Flash (high) (Claude Code) 0.454; GPT-5.6 Sol (Codex) 0.339.0.000.250.500.751.000153045607590105120135Elapsed time (min)Score SSonnet 5.5 · mini-SWE-agent: final S = 0.487 at 82 minGPT-6.1 Sol · mini-SWE-agent: final S = 0.494 at 121 minGPT-5.6 Sol · Codex: final S = 0.339 at 121 minGLM 5.3 Flash (high) · Claude Code: final S = 0.454 at 115 minSonnet 5.5 (high) · Claude Code: final S = 0.465 at 79 minKimi K3.1 Pro · Claude Code: final S = 0.473 at 121 minMuse Wonder (high) · Muse Code: final S = 0.480 at 47 minOpus 5.5 · Claude Code: final S = 0.483 at 56 minGLM 5.3 Flash (max) · mini-SWE-agent: final S = 0.484 at 87 minFable 5.1 · Claude Code: final S = 0.494 at 73 minGPT-6.1 Sol · Codex: final S = 0.498 at 121 minGemini 4 Argon · mini-SWE-agent: final S = 0.504 at 121 minGPT-6 Astra · Codex: final S = 0.508 at 121 min

Showing 13 of 13 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. 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

  • MZI-Spec-8 · Codex · GPT-6 Astra

    S = 0.508 · scale bar 500 µm

    Fit
    GPT-6 AstraCodexS = 0.51

  • MZI-Spec-8 · mini-SWE-agent · Gemini-4-Argon

    S = 0.504 · scale bar 500 µm

    Fit
    Gemini 4 Argonmini-SWE-agentS = 0.50

  • MZI-Spec-8 · Codex · GPT-6.1-sol

    S = 0.498 · scale bar 500 µm

    Fit
    GPT-6.1 SolCodexS = 0.50

  • MZI-Spec-8 · mini-SWE-agent · GPT-6.1-sol

    S = 0.494 · scale bar 500 µm

    Fit
    GPT-6.1 Solmini-SWE-agentS = 0.49

  • MZI-Spec-8 · Claude Code · Claude Fable 5.1

    S = 0.494 · scale bar 500 µm

    Fit
    Fable 5.1Claude CodeS = 0.49

  • MZI-Spec-8 · mini-SWE-agent · Claude Sonnet 5.5

    S = 0.487 · scale bar 500 µm

    Fit
    Sonnet 5.5mini-SWE-agentS = 0.49

  • MZI-Spec-8 · mini-SWE-agent · GLM 5.3 Flash

    S = 0.484 · scale bar 500 µm

    Fit
    GLM 5.3 Flash (max)mini-SWE-agentS = 0.48

  • MZI-Spec-8 · Claude Code · Claude Opus 5.5

    S = 0.483 · scale bar 500 µm

    Fit
    Opus 5.5Claude CodeS = 0.48

  • MZI-Spec-8 · Muse Code · Muse Wonder

    S = 0.480 · scale bar 500 µm

    Fit
    Muse Wonder (high)Muse CodeS = 0.48

  • MZI-Spec-8 · Claude Code · Kimi K3.1 Pro

    S = 0.473 · scale bar 500 µm

    Fit
    Kimi K3.1 ProClaude CodeS = 0.47

  • MZI-Spec-8 · Claude Code · Claude Sonnet 5.5

    S = 0.465 · scale bar 500 µm

    Fit
    Sonnet 5.5 (high)Claude CodeS = 0.47

  • MZI-Spec-8 · Claude Code · GLM 5.3 Flash

    S = 0.454 · scale bar 500 µm

    Fit
    GLM 5.3 Flash (high)Claude CodeS = 0.45

  • MZI-Spec-8 · Codex · GPT-5.6-sol

    S = 0.339 · scale bar 500 µm

    Fit
    GPT-5.6 SolCodexS = 0.34