--- title: "claude-code \u00b7 Opus 4.7 | Pelican benchmark" url: https://evaljetty.com/pelicans/runs/agents/claude-opus.html description: "claude-code \u00b7 Opus 4.7 (claude-opus-4-7) drawing a pelican riding a bicycle: every hill-climb run with self-scores, external judge scores and Jetty trajectories." updated: 2026-09-28 publisher: Jetty (https://jetty.io) --- 02 — Runs · v1 · jettyio · auto-evolved runbook (weakest-axis targeting) # claude-code claude-opus-4-7 Same agent as the reference, model swapped to Opus. Cleaner anatomy, plateaued at 36/40 on its own rubric. Judge, best round 28.7/40 round v3 (agent's pick) Self-score 36/40 inflation +7.3 Runs 10 not recorded (v1) ![claude-code · Opus 4.7 best round](https://evaljetty.com/pelicans/img/claude-opus/v3.svg) best round v3 · d4a107dc ### Self-score vs judge, by axis Best round v3, 0–10 per axis agent self-scoreexternal judge **Judge:** The pelican reads clearly, with a big orange pouched beak, a curved neck and white plumage, and the scene looks polished with sky, sun and grass. The wing reaching to the handlebars and the feet on the pedals sell the riding pose, but the bike frame is odd: the down tube runs to the front hub instead of the head tube, the pedals and cranks are ambiguous, and the wheels are cropped at the bottom. ### Score by round Out of 40. Each round starts from the previous round's best SVG. Dashed line: the 28/40 pass threshold. self-scorejudgepass ≥ 28 ## 10 drawings, *oldest first.* Click a drawing to open it in the head-to-head viewer. The outlined card is the agent's best round. Verdicts are derived (code checks on the stored SVG and judge ≥ 28/40); see [Runs](https://evaljetty.com/pelicans/runs/index.html#all-runs). [![claude-code · Opus 4.7 round v1](https://evaljetty.com/pelicans/img/claude-opus/v1.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#claude-opus/v1) **v1** passjudge **29** · self 33 c3121188[runbook](https://evaljetty.com/pelicans/author/versions.html#claude-opus/v1) [![claude-code · Opus 4.7 round v2](https://evaljetty.com/pelicans/img/claude-opus/v2.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#claude-opus/v2) **v2** passjudge **29.3** · self 35 cb2a7f37[runbook](https://evaljetty.com/pelicans/author/versions.html#claude-opus/v2) [![claude-code · Opus 4.7 round v3](https://evaljetty.com/pelicans/img/claude-opus/v3.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#claude-opus/v3) **v3** · best passjudge **28.7** · self 36 d4a107dc[runbook](https://evaljetty.com/pelicans/author/versions.html#claude-opus/v3) [![claude-code · Opus 4.7 round v4](https://evaljetty.com/pelicans/img/claude-opus/v4.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#claude-opus/v4) **v4** passjudge **29.7** · self 32 8ec2e90c[runbook](https://evaljetty.com/pelicans/author/versions.html#claude-opus/v4) [![claude-code · Opus 4.7 round v5](https://evaljetty.com/pelicans/img/claude-opus/v5.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#claude-opus/v5) **v5** passjudge **28** · self 34.5 4160dcf1[runbook](https://evaljetty.com/pelicans/author/versions.html#claude-opus/v5) [![claude-code · Opus 4.7 round v6](https://evaljetty.com/pelicans/img/claude-opus/v6.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#claude-opus/v6) **v6** failjudge **27.3** · self 36 4848f8a3[runbook](https://evaljetty.com/pelicans/author/versions.html#claude-opus/v6) [![claude-code · Opus 4.7 round v7](https://evaljetty.com/pelicans/img/claude-opus/v7.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#claude-opus/v7) **v7** failjudge **27.5** · self 33 df18c780[runbook](https://evaljetty.com/pelicans/author/versions.html#claude-opus/v7) [![claude-code · Opus 4.7 round v8](https://evaljetty.com/pelicans/img/claude-opus/v8.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#claude-opus/v8) **v8** failjudge **27** · self 33 20f46c4d[runbook](https://evaljetty.com/pelicans/author/versions.html#claude-opus/v8) [![claude-code · Opus 4.7 round v9](https://evaljetty.com/pelicans/img/claude-opus/v9.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#claude-opus/v9) **v9** passjudge **28** · self 33.5 4897cbca[runbook](https://evaljetty.com/pelicans/author/versions.html#claude-opus/v9) [![claude-code · Opus 4.7 round v10](https://evaljetty.com/pelicans/img/claude-opus/v10.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#claude-opus/v10) **v10** failjudge **27.7** · self 34 38634df1[runbook](https://evaljetty.com/pelicans/author/versions.html#claude-opus/v10) ### Scores per round | Round | Pelican | Bicycle | Composition | Polish | Self total | Judge total | Verdict | Judge spread | Cost | Minutes | Trajectory | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | v1 | 7self 8 | 7.5self 8 | 7.5self 9 | 7self 8 | 33 | 29 | pass | 0 | – | – | c3121188 | | v2 | 8self 9 | 6.5self 9 | 7.5self 9 | 7.3self 8 | 35 | 29.3 | pass | 0.5 | – | – | cb2a7f37 | | v3 | 7.5self 9 | 6.3self 9 | 7.7self 9 | 7.2self 9 | 36 | 28.7 | pass | 1.5 | – | – | d4a107dc | | v4 | 7.5self 8 | 7self 8 | 7.7self 8 | 7.5self 8 | 32 | 29.7 | pass | 0.5 | – | – | 8ec2e90c | | v5 | 7.5self 8.5 | 6self 9 | 7.5self 8.5 | 7self 8.5 | 34.5 | 28 | pass | 0 | – | – | 4160dcf1 | | v6 | 7.5self 9.5 | 5.8self 9 | 7self 8.5 | 7self 9 | 36 | 27.3 | fail | 0.5 | – | – | 4848f8a3 | | v7 | 7.3self 8 | 5.8self 8 | 7.3self 10 | 7self 7 | 33 | 27.5 | fail | 1 | – | – | df18c780 | | v8 | 7.3self 8 | 5.7self 8 | 7self 8 | 7self 9 | 33 | 27 | fail | 1 | – | – | 20f46c4d | | v9 | 7.7self 9 | 5.8self 8 | 7.5self 8 | 7self 8.5 | 33.5 | 28 | pass | 0 | – | – | 4897cbca | | v10 | 7.5self 8.5 | 6.2self 9 | 7self 8.5 | 7self 8 | 34 | 27.7 | fail | 0.5 | – | – | 38634df1 | --- Source: https://evaljetty.com/pelicans/runs/agents/claude-opus.html · Evals by Jetty · Run your own evals: https://jetty.io/?utm_source=evaljetty&utm_medium=llms