Skip to content
Leaderboard / Google

Gemini 3.8 Flash

Google High CLOSED Proprietary
Verdict

Gemini 3.8 Flash is a strong all-rounder: #28 of 406 by SIscope score. Strongest at chat quality, weaker at agents & tools. Mid-range on price.

SIscope score
87
#28 of 406
Human-vote rank
#9
of 419 · score 1499
Price per 1M tokens, blended
$1.65
$0.82 in · $4.13 out
Context window
1M
1,048,576 tokens

Shape

Each spoke is a category, 0 at the centre, 100 at the edge.

CodingAgentsReasoningMathChatInstructions

Category scores

Percentiles: 90 means better than about 90% of tested models.

Coding86
Agents & tools71
Reasoning & knowledge96
Math88
Chat quality98
Following instructions97

What goes into each category · Compare with DeepSeek V4.1 Flash

Human-vote score over time

Text arena, recent updates.

149014981505Oct 15: 1498.9 (rank #8)Oct 20: 1498.9 (rank #8)Oct 22: 1498.3 (rank #8)Oct 28: 1498.5 (rank #9)Oct 15Oct 28

Coding votes over time

WebDev arena, recent updates.

157515831590Oct 16: 1585.2 (rank #31)Oct 18: 1585.7 (rank #33)Oct 19: 1584.7 (rank #35)Oct 22: 1585.3 (rank #34)Oct 23: 1584.9 (rank #36)Oct 27: 1583.6 (rank #36)Oct 28: 1583.3 (rank #37)Oct 16Oct 28

Independent test runs

Run by Epoch AI on the same harness for every model. Bars show the share of questions answered correctly.

GPQA Diamond95%
PhD-level science multiple choice questions
Better than 99% of 333 tested models. Run Sep 2, 2026.
SimpleQA Verified70%
Short factual questions; rewards not making things up
Better than 92% of 100 tested models. Run Sep 2, 2026.
OTIS Mock AIME 2024-2599%
Competition maths problems in the style of AIME
Better than 92% of 313 tested models. Run Sep 2, 2026.
FrontierMath Tiers 1-368%
Very hard, unpublished research-style maths problems
Better than 74% of 122 tested models. Run Sep 2, 2026.

Human-vote arenas

People compare two anonymous answers and pick the better one. Scores work like chess ratings.

ArenaRankScoreRange (95%)VotesUpdated
Text arena
Everyday chat, blind human votes
#9 of 41914991494–150347,803Oct 28, 2026
WebDev arena
Building web apps, blind human votes
#37 of 15215831576–159012,342Oct 28, 2026
Vision arena
Questions about images, blind human votes
#7 of 16713151306–13254,645Oct 19, 2026
Agent arena
Real multi-step agent sessions
#17 of 572.6%1.7–3.4%52,949Nov 2, 2026

Source: Arena (formerly LMArena) leaderboard dataset, CC BY 4.0. Fetched Nov 4, 2026.

Price

Input
$0.82
per 1M tokens
Output
$4.13
per 1M tokens
Blended
$1.65
3 in : 1 out, per 1M
Context window
1M
max tokens per request
Speed
Coming soon
our own tests

List price from the OpenRouter models API (google/gemini-3.8-flash), Nov 4, 2026. Providers can charge differently.