Gemini 3.5 Flash vs Gemini 3 Flash Preview
Wins 5 of 8 areas
Coding · Agents · Reasoning · Images and charts · Math
Wins 2 of 8 areas
Long documents · Following instructions
Gemini 3.5 Flash is the stronger all-rounder.Gemini 3 Flash Preview is cheaper and better at long documents.
Scores updated · 45 tests both models report · How we compare
Where each one wins
Tests won in each of the eight areas we test. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software50Gemini 3.5 Flash5 of 5 tests
- ReasoningHard problems that need careful thinking50Gemini 3.5 Flash5 of 5 tests
- AgentsCarrying out multi-step tasks on its own40Gemini 3.5 Flash4 of 5 tests · 1 tie
- Images and chartsUnderstanding pictures, charts and video30Gemini 3.5 Flash3 of 3 tests
- MathCompetition and research-level math31Gemini 3.5 Flash3 of 4 tests
- Long documentsFinding answers in very long texts02Gemini 3 Flash Preview2 of 2 tests
- Following instructionsDoing exactly what it is asked01Gemini 3 Flash Preview1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each · 1 tie
Following instructions rests on a single test.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
Gemini 3 Flash Preview costs 67% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Gemini 3.5 Flash pulls ahead
- Abstract visual puzzles that people can solveARC-AGI-2+38.5points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+30.8points ahead
- Multi-step tasks using many real software toolsMCP Atlas+26.2points ahead
Where Gemini 3 Flash Preview pulls ahead
- Finds one of eight look-alike replies in a long chatMRCR v2 8 needle 128k (average)+40.6points ahead
- Reasons across sets of long documentsAA-LCR+4.7points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+2points ahead
Every test, side by side
All 45 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGemini 3.5 Flash
- Terminal-Bench 2.1Gemini 3.5 Flash by 20.778.758+20.7
- LMArena · WebDevGemini 3.5 Flash by 60 rating points14991439+60 rating
- SWE-bench ProGemini 3.5 Flash by 5.555.149.6+5.5
- SciCodeGemini 3.5 Flash by 3.353.950.6+3.3
- Terminal-Bench HardGemini 3.5 Flash by 2.340.938.6+2.3
AgentsGemini 3.5 Flash
- MCP AtlasGemini 3.5 Flash by 26.283.657.4+26.2
- AA ApexAgentsGemini 3.5 Flash by 19.447.127.7+19.4
- OSWorld-VerifiedGemini 3.5 Flash by 13.378.465.1+13.3
- τ-Bench V3 · BankingGemini 3.5 Flash by 11.432.220.8+11.4
- GDPValtie35.235.2tie
ReasoningGemini 3.5 Flash
- ARC-AGI-2Gemini 3.5 Flash by 38.572.133.6+38.5
- SimpleBenchGemini 3.5 Flash by 15.676.761.1+15.6
- Humanity's Last ExamGemini 3.5 Flash by 6.142.736.6+6.1
- CritPtGemini 3.5 Flash by 4.513.18.6+4.5
- GPQA DiamondGemini 3.5 Flash by 2.492.289.8+2.4
FactsEven
- AA-Omniscience · Non-hallucinationGemini 3.5 Flash by 30.837.87+30.8
- AA-Omniscience · AccuracyGemini 3 Flash Preview by 251.453.4+2
- SimpleQA Verifiedtie66.266.8tie
Images and chartsGemini 3.5 Flash
- MMMU-ProGemini 3.5 Flash by 4.484.379.9+4.4
- CharXiv (RQ)Gemini 3.5 Flash by 3.984.280.3+3.9
- LMArena · VisionGemini 3.5 Flash by 25 rating points13101285+25 rating
MathGemini 3.5 Flash
- FrontierMath Tiers 1-3 (v2)Gemini 3.5 Flash by 11.662.851.2+11.6
- FrontierMath Tier 4Gemini 3.5 Flash by 9.726.817.1+9.7
- HMMT Feb. 2026Gemini 3.5 Flash by 6.195.589.4+6.1
- AIME 2026Gemini 3 Flash Preview by 1.79596.7+1.7
Long documentsGemini 3 Flash Preview
- MRCR v2 8 needle 128k (average)Gemini 3 Flash Preview by 40.626.667.2+40.6
- AA-LCRGemini 3 Flash Preview by 4.773.378+4.7
Following instructionsGemini 3 Flash Preview
- IFBenchGemini 3 Flash Preview by 1.776.378+1.7
Other results17 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- GDPval-AA (Elo)Gemini 3.5 Flash by 452 rating points16561204+452 rating
- Blueprint-Bench 2Gemini 3.5 Flash by 33.633.60+33.6
- Terminal-Bench 2.0Gemini 3.5 Flash by 28.676.247.6+28.6
- Artificial Analysis Coding IndexGemini 3.5 Flash by 27.570.142.6+27.5
- AA Agentic IndexGemini 3 Flash Preview by 22.427.349.7+22.4
- Finance Agent v2Gemini 3.5 Flash by 15.357.942.5+15.3
- τ²-Bench Telecom (AA run)Gemini 3.5 Flash by 14.995.380.4+14.9
- AA-OmniscienceGemini 3.5 Flash by 11.121.210.1+11.1
- frontiermath_tier_4_v1Gemini 3.5 Flash by 10.414.64.2+10.4
- ARC-AGI-1Gemini 3.5 Flash by 7.892.584.7+7.8
- ToolathlonGemini 3.5 Flash by 7.156.549.4+7.1
- Humanity’s Last ExamGemini 3.5 Flash by 6.540.233.7+6.5
- AA IntelligenceGemini 3.5 Flash by 6.332.626.3+6.3
- mrcr_v2_8needle_1m_pointwiseGemini 3 Flash Preview by 4.222.126.3+4.2
- matharena_visual_math_overallGemini 3.5 Flash by 4.189.985.8+4.1
- LiveBenchGemini 3.5 Flash by 2.67572.4+2.6
- Legal Agent Benchmarktie0.80tie
Questions people ask
Which is better, Gemini 3.5 Flash or Gemini 3 Flash Preview?
Gemini 3.5 Flash wins five of the eight areas we test: coding, agents, reasoning, images and charts and math. Gemini 3 Flash Preview wins long documents and following instructions, and costs 67% less. They are level on facts.
Which is better for coding?
Gemini 3.5 Flash. It wins 5 of the 5 coding tests both models report; Gemini 3 Flash Preview wins none.
Which is cheaper?
Gemini 3.5 Flash costs $1.50 per million input tokens and $9.00 per million output tokens; Gemini 3 Flash Preview costs $0.50 and $3.00. That makes Gemini 3 Flash Preview about 67% cheaper for the same work.
How do you compare the two?
We use the 45 benchmark tests both models have published scores on. The verdict counts the 28 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 17 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.