Gemini 3.7 Flash vs Gemini 3.8 Flash
Wins 3 of 8 areas
Coding · Images and charts · Math
Wins 4 of 8 areas
Agents · Reasoning · Facts · Following instructions
Gemini 3.8 Flash wins more areas, narrowly.Gemini 3.7 Flash is better at math.
Scores updated · 41 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.
- ReasoningHard problems that need careful thinking03Gemini 3.8 Flash3 of 5 tests · 2 ties
- AgentsCarrying out multi-step tasks on its own03Gemini 3.8 Flash3 of 3 tests
- FactsGetting facts right instead of making them up01Gemini 3.8 Flash1 of 3 tests · 2 ties
- Following instructionsDoing exactly what it is asked01Gemini 3.8 Flash1 of 1 test
- MathCompetition and research-level math30Gemini 3.7 Flash3 of 3 tests
- CodingWriting and fixing software21Gemini 3.7 Flash2 of 5 tests · 2 ties
- Images and chartsUnderstanding pictures, charts and video10Gemini 3.7 Flash1 of 3 tests · 2 ties
- Long documentsFinding answers in very long texts00Even0 each · 1 tie
Long documents and following instructions rest on a single test each.
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.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Gemini 3.8 Flash pulls ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+12.1points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+9.3points ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+6.1points ahead
Where Gemini 3.7 Flash pulls ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+14.6points ahead
- Programming problems, refreshed regularlyLiveBench · Coding+6.4points ahead
- Fixes real GitHub issues, working as an agentLiveBench · Agentic Coding+4.1points ahead
Every test, side by side
All 41 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGemini 3.7 Flash
- LiveBench · CodingGemini 3.7 Flash by 6.478.972.5+6.4
- LiveBench · Agentic CodingGemini 3.7 Flash by 4.158.354.2+4.1
- Terminal-Bench 2.1Gemini 3.8 Flash by 1.885.887.6+1.8
- SciCodetie57.256.6tie
- LMArena · WebDevtie15901586tie
AgentsGemini 3.8 Flash
- τ-Bench V3 · BankingGemini 3.8 Flash by 12.132.844.9+12.1
- Terminal-Bench 4.0Gemini 3.8 Flash by 6.113.619.7+6.1
- GDPValGemini 3.8 Flash by 2.244.646.8+2.2
ReasoningGemini 3.8 Flash
- ARC-AGI-2Gemini 3.8 Flash by 4.684.689.2+4.6
- CritPtGemini 3.8 Flash by 414.318.3+4
- LiveBench · ReasoningGemini 3.8 Flash by 1.587.889.3+1.5
- GPQA Diamondtie94.595.3tie
- Humanity's Last Examtie47.947.8tie
FactsGemini 3.8 Flash
- AA-Omniscience · Non-hallucinationGemini 3.8 Flash by 9.335.544.8+9.3
- AA-Omniscience · Accuracytie55.354.6tie
- SimpleQA Verifiedtie69.269.7tie
Images and chartsGemini 3.7 Flash
- CharXiv (RQ)Gemini 3.7 Flash by 2.588.786.2+2.5
- LMArena · Visiontie13161314tie
- MMMU-Protie85.585.6tie
MathGemini 3.7 Flash
- FrontierMath Tier 4Gemini 3.7 Flash by 14.636.621.9+14.6
- FrontierMath Tiers 1-3 (v2)Gemini 3.7 Flash by 3.271.668.4+3.2
- LiveBench · MathematicsGemini 3.7 Flash by 1.993.591.6+1.9
Following instructionsGemini 3.8 Flash
- LiveBench · Instruction FollowingGemini 3.8 Flash by 1.579.981.4+1.5
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.
- livebench_data_analysisGemini 3.7 Flash by 146854+14
- OSWorld 2.0Gemini 3.8 Flash by 11.147.959+11.1
- DeepSWE 1.1Gemini 3.8 Flash by 8.465.373.7+8.4
- OSWorld 2.0 (partial)Gemini 3.8 Flash by 8.450.659+8.4
- Terminal-Bench 4.0Gemini 3.8 Flash by 7.911.219.1+7.9
- GDPval-AA v2 EloGemini 3.8 Flash by 63 rating points14821545+63 rating
- AA Agentic IndexGemini 3.8 Flash by 4.736.441.1+4.7
- AA-OmniscienceGemini 3.8 Flash by 3.126.529.6+3.1
- ARC-AGI-1Gemini 3.8 Flash by 395.598.5+3
- livebench_languageGemini 3.8 Flash by 2.385.587.8+2.3
- GDPval-AA v2Gemini 3.8 Flash by 20 rating points15251545+20 rating
- AA IntelligenceGemini 3.8 Flash by 1.839.140.9+1.8
- LVBenchGemini 3.8 Flash by 1.785.487.1+1.7
- HLE-VerifiedGemini 3.8 Flash by 1.353.654.9+1.3
- GDP (Surge AI)Gemini 3.8 Flash by 13435+1
- GDP.PDF (All pass rate)Gemini 3.8 Flash by 13435+1
- Artificial Analysis Coding Indextie76.176.3tie
Questions people ask
Which is better, Gemini 3.7 Flash or Gemini 3.8 Flash?
Gemini 3.8 Flash wins four of the eight areas we test: agents, reasoning, facts and following instructions. Gemini 3.7 Flash wins coding, images and charts and math. They are level on long documents.
Which is better for coding?
Gemini 3.7 Flash. It wins 2 of the 5 coding tests both models report; Gemini 3.8 Flash wins 1, and 2 are ties.
Which is cheaper?
Gemini 3.7 Flash costs $0.75 per million input tokens and $3.75 per million output tokens; Gemini 3.8 Flash costs $0.75 and $3.75. For a million tokens read plus a million written, they cost about the same.
How do you compare the two?
We use the 41 benchmark tests both models have published scores on. The verdict counts the 24 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.