Noometry

Google, proprietary

Gemini 3 Flash Preview

Gemini 3 Flash Preview by Google ranks 40th of 354 ranked models on the Noometry Index as of October 2026, with a score of 52.3. Its strongest category is multimodal, where it ranks 16th. API pricing starts at $0.50 per million input tokens and $3 per million output tokens, with a 1.05M-token context window.

Last verified

Noometry rank
#40 of 354
Index score
52.3
Evidence
Confirmed 59 results
Provider
Google
Released
December 17, 2025
Weights
Proprietary
Reasoning
Yes
Context window
1.05M
Max output
66K
Input price
$0.50 / M
Output price
$3 / M
Blended price
$1.13 / M
Output speed
Not measured
Value
#100 of 219
Knowledge cutoff
January 2025
Input
text, image, video, audio, pdf

Category scores

Each category score combines every public result we have in that category.

  1. Coding 50.9
  2. Agentic & Tool Use 38.7
  3. Reasoning 49.2
  4. Math 51.7
  5. Knowledge 58.8
  6. Multimodal 45.5
  7. Multilingual 55.7
  8. Instruction Following 75.7
  9. Long Context 44.4
  10. Writing & Preference 65.5
CategoryScoreRankResults
Coding50.9#427
Agentic & Tool Use38.7#298
Reasoning49.2#379
Math51.7#556
Knowledge58.8#334
Multimodal45.5#164
Multilingual55.7#271
Instruction Following75.7#561
Long Context44.4#671
Writing & Preference65.5#453

Strengths and weaknesses

Categories where Gemini 3 Flash Preview places highest and lowest among the models ranked in each, with its score against that category's median.

Strongest categories

CategoryScorevs medianRank
Multilingual55.7+8.3#27 of 297, top 10%
Knowledge58.8+21.5#33 of 314, top 11%
Reasoning49.2+25.6#37 of 350, top 11%

Weakest categories

CategoryScorevs medianRank
Long Context44.4+3.5#67 of 296, top 23%
Agentic & Tool Use38.7+8.4#29 of 154, top 19%
Instruction Following75.7+4.4#56 of 305, top 19%

Closest competitors

The models ranked just above and below Gemini 3 Flash Preview. When scores are this close, price and speed are often the better way to choose.

ModelRankScoreBlended $/MSpeed
GPT-6 Luna#3653.3$0.20—Compare
Grok 4.7#3753.1$3—Compare
DeepSeek V4.1 Flash#3852.8$0.26—Compare
GPT-5.2 Pro#3952.3$57.75—Compare
GLM-5.3-Flash#4151.8$0.24—Compare
Qwen3.7 Max#4251.5$3.75—Compare
Qwen3.6 Max Preview#4351.5$2.92—Compare
GLM-5.2#4451.1$2.1523Compare

Sponsored placements are available on pages like this one. Advertise on Noometry

Benchmark results

Every published result we track, with its source. Bold rows are the ones used for ranking; where several exist we prefer independent runs over self-reported numbers.

Coding

BenchmarkScorePositionSettingSourceDate
SWE-bench Verified75.4%#13 of 32, top 41%Epoch AI2026-02-18
SWE-bench Verified (bash only)75.8%#2 of 39, top 6%highSWE-bench2026-02-17
LMArena WebDev1385LMArena2026-10-08
LMArena WebDev1439#59 of 113, top 53%LMArena2026-10-08
SWE-bench Multilingual72.7%Best of 13SWE-bench2026-02-13
GSO9.8%#17 of 31, top 55%Epoch AI
WeirdML61.6%#27 of 119, top 23%Epoch AI
LMArena Coding1460#71 of 294, top 25%LMArena2026-10-08
LMArena Coding1436LMArena2026-10-08
ALE-Bench1,367#20 of 105, top 20%Epoch AI

Agentic & Tool Use

BenchmarkScorePositionSettingSourceDate
Terminal-Bench64.3%#10 of 41, top 25%Epoch AI
τ²-bench Airline82.5%#3 of 7, top 43%highτ²-bench2026-03-02
τ²-bench Banking27.3%#15 of 26, top 58%highτ²-bench2026-03-02
τ²-bench Retail76.8%#4 of 7, top 58%highτ²-bench2026-03-02
τ²-bench Telecom91.2%#3 of 7, top 43%highτ²-bench2026-03-02
DeepResearch Bench47.9%highEpoch AI
DeepResearch Bench49.8%#7 of 24, top 30%lowEpoch AI
DeepResearch Bench49%minimalEpoch AI
BALROG48.1%#7 of 35, top 20%Epoch AI
GDP.pdf10%#34 of 36, top 95%Epoch AI
GDP.pdf10%highEpoch AI
LMArena Search1198#15 of 32, top 47%LMArena2026-08-24
Vending-Bench 23,635#39 of 60, top 65%Epoch AI

Reasoning

BenchmarkScorePositionSettingSourceDate
ARC-AGI-233.6%#39 of 83, top 47%Epoch AI
SimpleBench61.1%#22 of 77, top 29%Epoch AI
NYT Connections (extended)83.1%#32 of 91, top 36%Lech Mazur benchmarks
ARC-AGI-184.7%#36 of 83, top 44%Epoch AI
Chess Puzzles38%Epoch AI2025-12-17
Chess Puzzles40%#19 of 129, top 15%highEpoch AI2026-08-06
LMArena Hard Prompts1438LMArena2026-10-08
LMArena Hard Prompts1465#41 of 297, top 14%LMArena2026-10-08
Mystery Game Puzzles20%highEpoch AI2026-08-05
Mystery Game Puzzles26%#31 of 74, top 42%lowEpoch AI2026-08-05
Mystery Game Puzzles25%minimalEpoch AI2026-08-05
DTBench89.1%#44 of 151, top 30%highEpoch AI
LMCA43.1%#40 of 125, top 32%highEpoch AI
Epoch Capabilities Index151.8#47 of 213, top 23%Epoch AI2025-12-17
ForecastBench58.5#47 of 72, top 66%Epoch AI

Math

BenchmarkScorePositionSettingSourceDate
FrontierMath (Tiers 1-3)51.2%#46 of 81, top 57%Epoch AI2026-06-11
FrontierMath Tier 417.1%#44 of 63, top 70%Epoch AI2026-06-11
MathArena Final-Answer Competitions67.6%#15 of 29, top 52%MathArena
OTIS Mock AIME 2024-202592.8%Epoch AI2025-12-17
OTIS Mock AIME 2024-202595.6%#33 of 173, top 20%highEpoch AI2026-08-06
ProofBench15%#57 of 77, top 75%Epoch AI
LMArena Math1448LMArena2026-10-08
LMArena Math1473#37 of 285, top 13%LMArena2026-10-08
FrontierMath (Feb 2025 set)35.6%#14 of 68, top 21%Epoch AI2025-12-17
FrontierMath Tier 4 (v1)4.2%#33 of 55, top 60%Epoch AI2025-12-17

Knowledge

BenchmarkScorePositionSettingSourceDate
GPQA Diamond83.2%Epoch AI2025-12-17
GPQA Diamond89.4%#43 of 186, top 24%highEpoch AI2026-08-06
SimpleQA Verified66.8%#10 of 77, top 13%highEpoch AI2026-08-27
Vectara Hallucination Rate (lower is better)13.5%#81 of 96, top 85%Vectara Hallucination Leaderboard
LMArena Expert1419LMArena2026-10-08
LMArena Expert1462#59 of 273, top 22%LMArena2026-10-08

Multimodal

BenchmarkScorePositionSettingSourceDate
LMArena Vision1285#25 of 122, top 21%LMArena2026-10-09
LMArena Vision1266LMArena2026-10-09
GeoBench88%Best of 25Epoch AI
VPCT72.6%#3 of 24, top 13%Epoch AI
Blueprint-Bench 20%#28 of 31, top 91%Epoch AI
LMArena Document1413#35 of 38, top 93%LMArena2026-09-13

Multilingual

BenchmarkScorePositionSettingSourceDate
LMArena Non-English1458#27 of 297, top 10%LMArena2026-10-08
LMArena Non-English1438LMArena2026-10-08
LMArena Chinese1477LMArena2026-10-08
LMArena Chinese1511#36 of 285, top 13%LMArena2026-10-08
LMArena French1477#28 of 223, top 13%LMArena2026-10-08
LMArena French1445LMArena2026-10-08
LMArena German1497#9 of 231, top 4%LMArena2026-10-08
LMArena German1466LMArena2026-10-08
LMArena Japanese1489#10 of 211, top 5%LMArena2026-10-08
LMArena Japanese1424LMArena2026-10-08
LMArena Korean1417LMArena2026-10-08
LMArena Korean1443#23 of 213, top 11%LMArena2026-10-08
LMArena Russian1480#22 of 283, top 8%LMArena2026-10-08
LMArena Russian1454LMArena2026-10-08
LMArena Spanish1469#26 of 226, top 12%LMArena2026-10-08
LMArena Spanish1427LMArena2026-10-08

Instruction Following

BenchmarkScorePositionSettingSourceDate
LMArena Instruction Following1415LMArena2026-10-08
LMArena Instruction Following1437#52 of 298, top 18%LMArena2026-10-08

Long Context

BenchmarkScorePositionSettingSourceDate
LMArena Longer Query1429LMArena2026-10-08
LMArena Longer Query1452#50 of 291, top 18%LMArena2026-10-08

Writing & Preference

BenchmarkScorePositionSettingSourceDate
LMArena Text1442LMArena2026-10-08
LMArena Text1466#30 of 297, top 11%LMArena2026-10-08
LMArena Creative Writing1457#21 of 295, top 8%LMArena2026-10-08
LMArena Creative Writing1435LMArena2026-10-08
LMArena Multi-Turn1471#29 of 295, top 10%LMArena2026-10-08
LMArena Multi-Turn1455LMArena2026-10-08

API pricing by provider

RouteInput $/MOutput $/MCached input $/MChecked
google$0.50$3$0.052026-10-10
openrouter$0.50$3$0.052026-10-10
vertex$0.50$3$0.052026-10-10

All Google API prices → Estimate your cost →

Compare Gemini 3 Flash Preview

Other Google models

Frequently asked questions

How good is Gemini 3 Flash Preview?

Gemini 3 Flash Preview by Google ranks 40th of 354 ranked models on the Noometry Index as of October 2026, with a score of 52.3. Its strongest category is multimodal, where it ranks 16th. API pricing starts at $0.50 per million input tokens and $3 per million output tokens, with a 1.05M-token context window.

How much does Gemini 3 Flash Preview cost?

Gemini 3 Flash Preview costs $0.50 per million input tokens and $3 per million output tokens on Google's own API, with cached input at $0.05.

What is Gemini 3 Flash Preview's context window?

Gemini 3 Flash Preview accepts up to 1.05M tokens of input and can write up to 66K tokens in one response.

Is Gemini 3 Flash Preview open source?

No. Gemini 3 Flash Preview is proprietary and available only through Google's API and partner platforms.

What are Gemini 3 Flash Preview's strengths and weaknesses?

Relative to other ranked models, Gemini 3 Flash Preview places best in multilingual, knowledge, reasoning and lowest in long context, agentic & tool use, instruction following.

What is Gemini 3 Flash Preview best at?

Its best category is multimodal, where it ranks 16th on Noometry.