--- title: "gemini-cli \u00b7 Gemini 3.5 Flash | Pelican benchmark" url: https://evaljetty.com/pelicans/runs/agents/gemini-cli.html description: "gemini-cli \u00b7 Gemini 3.5 Flash (gemini-3.5-flash) 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) # gemini-cli gemini-3.5-flash Fastest v1 runner. Flash grades itself generously: 40/40 is its ceiling, not a verdict. Judge, best round 29.8/40 round v1 (agent's pick) Self-score 40/40 inflation +10.2 Runs 10 not recorded (v1) ![gemini-cli · Gemini 3.5 Flash best round](https://evaljetty.com/pelicans/img/gemini-cli/v1.svg) best round v1 · f5cd6a31 ### Self-score vs judge, by axis Best round v1, 0–10 per axis agent self-scoreexternal judge **Judge:** A white pelican with an orange beak and throat pouch sits on a well-rendered spoked bicycle against an attractive retro sunset, with wing reaching the handlebar and feet near the pedals. Defects: the frame geometry is odd (seat tube and top tube join awkwardly, the fork is a single straight tube, and the head tube is missing), the pouch is oversized and blob-like, one foot floats oddly by the rear wheel, and the chain routing is simplified. ### 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). [![gemini-cli · Gemini 3.5 Flash round v1](https://evaljetty.com/pelicans/img/gemini-cli/v1.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#gemini-cli/v1) **v1** · best passjudge **29.8** · self 40 f5cd6a31[runbook](https://evaljetty.com/pelicans/author/versions.html#gemini-cli/v1) [![gemini-cli · Gemini 3.5 Flash round v2](https://evaljetty.com/pelicans/img/gemini-cli/v2.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#gemini-cli/v2) **v2** passjudge **31** · self 38.5 c990e7c4[runbook](https://evaljetty.com/pelicans/author/versions.html#gemini-cli/v2) [![gemini-cli · Gemini 3.5 Flash round v3](https://evaljetty.com/pelicans/img/gemini-cli/v3.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#gemini-cli/v3) **v3** passjudge **29.3** · self 38.4 5d360721[runbook](https://evaljetty.com/pelicans/author/versions.html#gemini-cli/v3) [![gemini-cli · Gemini 3.5 Flash round v4](https://evaljetty.com/pelicans/img/gemini-cli/v4.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#gemini-cli/v4) **v4** passjudge **31.7** · self 40 4ed5de35[runbook](https://evaljetty.com/pelicans/author/versions.html#gemini-cli/v4) [![gemini-cli · Gemini 3.5 Flash round v5](https://evaljetty.com/pelicans/img/gemini-cli/v5.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#gemini-cli/v5) **v5** passjudge **32.3** · self 39.5 a652e4e3[runbook](https://evaljetty.com/pelicans/author/versions.html#gemini-cli/v5) [![gemini-cli · Gemini 3.5 Flash round v6](https://evaljetty.com/pelicans/img/gemini-cli/v6.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#gemini-cli/v6) **v6** passjudge **30.5** · self 39.5 eca90d64[runbook](https://evaljetty.com/pelicans/author/versions.html#gemini-cli/v6) [![gemini-cli · Gemini 3.5 Flash round v7](https://evaljetty.com/pelicans/img/gemini-cli/v7.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#gemini-cli/v7) **v7** passjudge **31** · self 38.7 f08684e3[runbook](https://evaljetty.com/pelicans/author/versions.html#gemini-cli/v7) [![gemini-cli · Gemini 3.5 Flash round v8](https://evaljetty.com/pelicans/img/gemini-cli/v8.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#gemini-cli/v8) **v8** passjudge **31.7** · self 40 ce2de30f[runbook](https://evaljetty.com/pelicans/author/versions.html#gemini-cli/v8) [![gemini-cli · Gemini 3.5 Flash round v9](https://evaljetty.com/pelicans/img/gemini-cli/v9.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#gemini-cli/v9) **v9** passjudge **32** · self 37.5 4e122677[runbook](https://evaljetty.com/pelicans/author/versions.html#gemini-cli/v9) [![gemini-cli · Gemini 3.5 Flash round v10](https://evaljetty.com/pelicans/img/gemini-cli/v10.svg)](https://evaljetty.com/pelicans/investigations/best-pelican.html#gemini-cli/v10) **v10** passjudge **31.2** · self 39.2 24091968[runbook](https://evaljetty.com/pelicans/author/versions.html#gemini-cli/v10) ### Scores per round | Round | Pelican | Bicycle | Composition | Polish | Self total | Judge total | Verdict | Judge spread | Cost | Minutes | Trajectory | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | v1 | 7.3self 10 | 7self 10 | 7.5self 10 | 8self 10 | 40 | 29.8 | pass | 0.5 | – | – | f5cd6a31 | | v2 | 8self 9.5 | 7.2self 9.5 | 7.8self 9.5 | 8self 10 | 38.5 | 31 | pass | 0 | – | – | c990e7c4 | | v3 | 7.5self 9.8 | 6.8self 9.5 | 7.3self 9.5 | 7.7self 9.6 | 38.4 | 29.3 | pass | 2 | – | – | 5d360721 | | v4 | 7.7self 10 | 8self 10 | 7.8self 10 | 8.2self 10 | 40 | 31.7 | pass | 2 | – | – | 4ed5de35 | | v5 | 7.8self 9.9 | 8.3self 9.8 | 7.8self 9.9 | 8.3self 9.9 | 39.5 | 32.3 | pass | 2 | – | – | a652e4e3 | | v6 | 7.5self 9.8 | 8self 9.9 | 7self 9.8 | 8self 10 | 39.5 | 30.5 | pass | 0 | – | – | eca90d64 | | v7 | 7.3self 9.7 | 8.2self 9.7 | 7.5self 9.6 | 8self 9.7 | 38.7 | 31 | pass | 1.5 | – | – | f08684e3 | | v8 | 7.7self 10 | 8self 10 | 8self 10 | 8self 10 | 40 | 31.7 | pass | 1.5 | – | – | ce2de30f | | v9 | 7.8self 9.5 | 8.3self 9 | 7.7self 9.5 | 8.2self 9.5 | 37.5 | 32 | pass | 2 | – | – | 4e122677 | | v10 | 7.7self 9.8 | 8.2self 9.8 | 7.3self 9.8 | 8self 9.8 | 39.2 | 31.2 | pass | 1.5 | – | – | 24091968 | --- Source: https://evaljetty.com/pelicans/runs/agents/gemini-cli.html · Evals by Jetty · Run your own evals: https://jetty.io/?utm_source=evaljetty&utm_medium=llms