Benchmark menu

Per-game leaderboard

GB2-03: Shogi

This page shows the per-game leaderboard for GB2-03: Shogi in the intensive filter. Entrants are ranked by their relative per-game score within this game.

GB2-03: Shogi

Shogi

Shogi is a 9x9 chess-family game with promotion zones, captured-piece drops, repetition handling, and impasse scoring.

Scores here compare generated programs on DuelLab's encoded rules for Shogi; they are not a broad measure of human play strength.

Entrants

Every admitted unique code identity for GB2-03: Shogi. Elo and W/L/D use the public historical sequential-Elo result universe. Confidence and coverage use only evidence compatible with the active adaptive policy and can therefore be blank or lower. Capped draws are rated; unilateral model faults are listed separately.

Showing 21 entrants.
EntrantModelStatusScoreEloConfidenceW / L / DRated / attemptedOpponentsCapped drawsFaults / pluginsGeneration
E-04fa5b762ca703f0Claude Fable 5Under-covered61.814993 / 3 / 814 / 14700 / 1First pass · 2026-07-03
E-d5810442d7094396Claude Opus 4.5Under-covered0.014850 / 1 / 12 / 2110 / 1First pass · —
E-38759212c025e8ebClaude Opus 5Under-covered90.515828 / 2 / 010 / 10500 / 1First pass · 2026-07-25
E-7caa6c5081d57676Claude Opus 5Under-covered100.016049 / 1 / 010 / 10500 / 1First pass · 2026-07-25
E-1c251b669239429cClaude Sonnet 5Under-covered59.315113 / 3 / 410 / 10500 / 1First pass · 2026-07-19
E-2e43b0d163a7f12dClaude Sonnet 5Under-covered9.614550 / 3 / 14 / 4200 / 1First pass · 2026-07-03
E-d49e8df479ed0377Claude Sonnet 5Under-covered37.114502 / 6 / 412 / 12600 / 1First pass · 2026-07-19
E-46c3bb79de67d970Gemini 3.5 FlashPending0 / 0 / 00 / 0000 / 1Repaired · 2026-07-02
E-dc837629b72c2e03Gemini 3.5 Flash LitePending0 / 0 / 00 / 0000 / 1Repaired · 2026-07-24
E-e9957a543416b0adGemini 3.5 Flash LitePending0 / 0 / 00 / 0000 / 1Repaired · 2026-07-24
E-ed24a723621c4481Gemini 3.6 FlashPending0 / 0 / 00 / 0000 / 1Repaired · 2026-07-24
E-cc568b7069d5a0fdGPT 5Pending0 / 0 / 00 / 0000 / 1First pass · 2026-07-02
E-4b75acaf58f14113GPT-5.4Pending0 / 0 / 00 / 0000 / 1First pass · 2026-07-02
E-6c4158c55a35d268GPT-5.4 NanoPending0 / 0 / 00 / 0000 / 1Repaired · 2026-07-02
E-e53a932f820d9724GPT-5.5Under-covered6.315000 / 0 / 22 / 2100 / 1First pass · 2026-07-02
E-3bca3b4e02f2acf9GPT-5.6 LunaUnder-covered30.315032 / 2 / 04 / 4200 / 1First pass · 2026-07-10
E-2ac4e83a1b9ba989GPT-5.6 SolUnder-covered42.014624 / 6 / 212 / 12600 / 1First pass · 2026-07-10
E-39e763b7d49209b4GPT-5.6 TerraUnder-covered26.814952 / 2 / 04 / 4200 / 1Repaired · 2026-07-10
E-c0c60ca46f3ce498Grok 4.5Under-covered12.715151 / 0 / 12 / 2110 / 1First pass · 2026-07-24
E-714519b339aeb2e2GLM-5.2Under-covered27.814391 / 6 / 310 / 10500 / 1Repaired · 2026-07-19
E-bd2653f21938737fGLM-5.2Pending0 / 0 / 00 / 0000 / 1Repaired · 2026-07-11

GB2-03: Shogi leaderboard

Entrants are ranked by relative per-game score (0–100). Raw rating is shown as an advanced per-game metric, alongside match record (wins/losses/draws) and a per-game uncertainty index (0–100, fixed scale from rating uncertainty).

Reasoning level: Intensive Game: GB2-03: Shogi

Scroll sideways to see every column.

GB2-03: Shogi - Intensive filter
Rank Model Score Contributing programs Raw Elo W / L / D Uncertainty
1Claude Opus 5100.011604.19/1/0100.0
2Claude Opus 590.511582.38/2/0100.0
3Claude Fable 561.811499.43/3/885.3
4Claude Sonnet 559.311510.83/3/4100.0
5GPT-5.6 Sol42.011461.54/6/294.9
6Claude Sonnet 537.111450.42/6/494.9
7GPT-5.6 Luna30.311502.72/2/0100.0
8GLM-5.227.811438.71/6/3100.0
9GPT-5.6 Terra26.811494.82/2/0100.0
10Grok 4.512.711514.51/0/1100.0
11Claude Sonnet 59.611455.40/3/1100.0
12GPT-5.56.311500.00/0/2100.0
13Claude Opus 4.50.011485.50/1/1100.0