PhysDesignBench
GitHub
Design task · v0

90SiN Bend-v0

90 Degree SiN Waveguide Bend Design

Design a low-loss 90-degree silicon-nitride waveguide bend at 1550 nm with a fixed Tidy3D setup, trusted cloud evaluation, and geometry DRC.

Runs
13
Scored finals
13 of 13
Best final
S = 0.76 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) 0.761; Opus 5.5 (Claude Code) 0.739; GPT-6 Astra (Codex) 0.719; Gemini 4 Argon (mini-SWE-agent) 0.702; Muse Wonder (high) (Muse Code) 0.674; GPT-6.1 Sol (Codex) 0.653; Kimi K3.1 Pro (Claude Code) 0.643; GPT-5.6 Sol (Codex) 0.636; GLM 5.3 Flash (max) (mini-SWE-agent) 0.621; Sonnet 5.5 (mini-SWE-agent) 0.612; Sonnet 5.5 (high) (Claude Code) 0.609; GPT-6.1 Sol (mini-SWE-agent) 0.596; GLM 5.3 Flash (high) (Claude Code) 0.518.0.000.250.500.751.000153045607590105120135Elapsed time (min)Score SGPT-6.1 Sol · mini-SWE-agent: final S = 0.596 at 121 minSonnet 5.5 · mini-SWE-agent: final S = 0.612 at 32 minGLM 5.3 Flash (high) · Claude Code: final S = 0.518 at 90 minSonnet 5.5 (high) · Claude Code: final S = 0.609 at 25 minGLM 5.3 Flash (max) · mini-SWE-agent: final S = 0.621 at 52 minGPT-5.6 Sol · Codex: final S = 0.636 at 121 minKimi K3.1 Pro · Claude Code: final S = 0.643 at 121 minGPT-6.1 Sol · Codex: final S = 0.653 at 121 minMuse Wonder (high) · Muse Code: final S = 0.674 at 39 minGemini 4 Argon · mini-SWE-agent: final S = 0.702 at 121 minGPT-6 Astra · Codex: final S = 0.719 at 121 minOpus 5.5 · Claude Code: final S = 0.739 at 112 minFable 5.1 · Claude Code: final S = 0.761 at 80 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 geometry.

This task’s evaluator keeps geometry images rather than GDS; each is shown on its own axes. Hatched layouts failed a validity gate and score zero. Select a layout to zoom.

  • 90SiN Bend-v0 · Claude Code · Claude Fable 5.1

    S = 0.761

    Fit
    Fable 5.1Claude CodeS = 0.76

  • 90SiN Bend-v0 · Claude Code · Claude Opus 5.5

    S = 0.739

    Fit
    Opus 5.5Claude CodeS = 0.74

  • 90SiN Bend-v0 · Codex · GPT-6 Astra

    S = 0.719

    Fit
    GPT-6 AstraCodexS = 0.72

  • 90SiN Bend-v0 · mini-SWE-agent · Gemini-4-Argon

    S = 0.702

    Fit
    Gemini 4 Argonmini-SWE-agentS = 0.70

  • 90SiN Bend-v0 · Muse Code · Muse Wonder

    S = 0.674

    Fit
    Muse Wonder (high)Muse CodeS = 0.67

  • 90SiN Bend-v0 · Codex · GPT-6.1-sol

    S = 0.653

    Fit
    GPT-6.1 SolCodexS = 0.65

  • 90SiN Bend-v0 · Claude Code · Kimi K3.1 Pro

    S = 0.643

    Fit
    Kimi K3.1 ProClaude CodeS = 0.64

  • 90SiN Bend-v0 · Codex · GPT-5.6-sol

    S = 0.636

    Fit
    GPT-5.6 SolCodexS = 0.64

  • 90SiN Bend-v0 · mini-SWE-agent · GLM 5.3 Flash

    S = 0.621

    Fit
    GLM 5.3 Flash (max)mini-SWE-agentS = 0.62

  • 90SiN Bend-v0 · mini-SWE-agent · Claude Sonnet 5.5

    S = 0.612

    Fit
    Sonnet 5.5mini-SWE-agentS = 0.61

  • 90SiN Bend-v0 · Claude Code · Claude Sonnet 5.5

    S = 0.609

    Fit
    Sonnet 5.5 (high)Claude CodeS = 0.61

  • 90SiN Bend-v0 · mini-SWE-agent · GPT-6.1-sol

    S = 0.596

    Fit
    GPT-6.1 Solmini-SWE-agentS = 0.60

  • 90SiN Bend-v0 · Claude Code · GLM 5.3 Flash

    S = 0.518

    Fit
    GLM 5.3 Flash (high)Claude CodeS = 0.52