Gemini 3.8 Flash

Google, closed weights, released Sep 2026, run at provider defaults through OpenRouter, served by Google AI Studio. It built the same five pages as every model on season 0.5 of the Rams Index, and Rams graded each one at desktop and phone size.

Rams Index, 0 to 100

55.5

95% interval 50.0–62.0

Rank: 12of 14 models by Index; the paired intervals allow 8 to 14
Cost per page: $0.056Generation only, list price on run day
Second judge: 69.6GPT-5.6 Terra places it 13th of 14. The Claude judge: 12th.
Grade sum: 22.55Seven 1–5 grades, out of 35
Pages judged: 1010 judge passes, 2 builds per brief
Defects per page: 5.0Tap targets under 24px on the phone (8 of 10)

Rank 12 of 14; the data allows anywhere from 8 to 14. Within noise of Claude Sonnet 5.5, Claude Opus 5.5, Claude Fable 5.1, Claude Fable 5, DeepSeek V4 Pro and Claude Opus 4.8: the gap to each sits inside the noise, so the board can't order them yet. Clearly behind GPT-6 Astra, Grok 4.7, GPT-6.1 Sol, GPT-6 Luna, Gemini 4 Argon, Muse Wonder and GPT-5.6 Sol. Not clearly ahead of any model.

Designer check pending · read Nov 6Season 0.5New this seasonProvisional: 1.0 judge passes per page, protocol 2

Grades by dimension

The seven grades behind the Index, 1 to 5, mean of desktop and phone. The tick is the mean across all 14 models.

Hierarchy3.1514 of 14
Typography3.2511 of 14
Spacing3.1011 of 14
Color3.358 of 14
Components3.3510 of 14
Composure3.0514 of 14
Craft3.3011 of 14

Bars start at 2.8 and end at 4.2, the range the board uses, so small differences show.

Index by brief

The same Index on each of the five briefs, and where the model places on that brief among all 14.

Landing64.38 of 14
Dashboard50.012 of 14
Settings50.010 of 14
Pricing50.010 of 14
Article63.414 of 14

Bars run from 40 to 90.

How it ran

Every model runs at its provider's default settings, through the same harness, with the same budget per page.

Effort: DefaultThe provider’s default. We send no effort or thinking setting
Time per page: 94sMedian over every build, broken ones included
Output tokens: 14.4kMedian per page, reasoning included
Reasoning tokens: 1.6kMedian per page, as the API reports them
Broken pages: 0 of 10Every build came back as a whole page
Criticals: 40%4 of 10 scored pages had a critical issue in their first Rams code review
Built through: OpenRouterServed by Google AI Studio

Best and worst page

Of its 10 judged pages, the one the judge graded highest and the one it graded lowest, with the judge's top findings on each, most severe first.

Best: Landing69.6
Index on this page, one judge pass. Brief mean 64.3.
- Subtext checkmarks misaligned: left edges staggerSerious
- Two-line hero headline split creates uneven tonal weightModerate
- Feature cards have uneven vertical rhythm between body copy and footer metricModerate
Worst: Pricing48.2
Index on this page, one judge pass. Brief mean 50.0.
- Comparison table redundantly labels Pro column 'Recommended'Serious
- Card vertical alignment is uneven across the three pricing tiersSerious
- Toggle billing labels misaligned verticallySerious

All 10 of its pages sit beside every other model's in the gallery.

Head to head

Gemini 3.8 Flash against every other model: the paired difference in Index points, and whether the 95% interval on it clears zero.

vs GPT-6 Astra-10.9Clearly behind
vs Grok 4.7-10.5Clearly behind
vs GPT-6.1 Sol-9.8Clearly behind
vs Claude Sonnet 5.5-6.2Inside the noise
vs GPT-6 Luna-6.2Clearly behind
vs Gemini 4 Argon-6.1Clearly behind
vs Muse Wonder-5.0Clearly behind
vs Claude Opus 5.5-4.8Inside the noise
vs GPT-5.6 Sol-3.4Clearly behind
vs Claude Fable 5.1-3.4Inside the noise
vs Claude Fable 5-2.1Inside the noise
vs DeepSeek V4 Pro+0.9Inside the noise
vs Claude Opus 4.8+2.7Inside the noise

The other models

Season 0.5 of the Rams Index, in board order, then the unranked.

Rams Index v0.5 · updated 2026-10-17Judged by Rams eyes · Claude Sonnet 4.6