Gemini 3.8 Flash vs Claude Opus 5
Wins 4 of 8 areas
Facts · Images and charts · Long documents · Following instructions
Wins 4 of 8 areas
Coding · Agents · Reasoning · Math
The two are evenly matched.Gemini 3.8 Flash is better at facts and images and charts; Claude Opus 5 at coding and reasoning.
Both rank among the ten best models we track in reasoning.
Scores updated · 44 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 software04Claude Opus 54 of 5 tests · 1 tie
- ReasoningHard problems that need careful thinking25Claude Opus 55 of 7 tests
- MathCompetition and research-level math03Claude Opus 53 of 3 tests
- AgentsCarrying out multi-step tasks on its own12Claude Opus 52 of 3 tests
- FactsGetting facts right instead of making them up21Gemini 3.8 Flash2 of 3 tests
- Images and chartsUnderstanding pictures, charts and video10Gemini 3.8 Flash1 of 3 tests · 2 ties
- Long documentsFinding answers in very long texts10Gemini 3.8 Flash1 of 1 test
- Following instructionsDoing exactly what it is asked10Gemini 3.8 Flash1 of 1 test
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.
Gemini 3.8 Flash costs 85% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Claude Opus 5 pulls ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+51.2points ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+29.3points ahead
- Solves new puzzle games it has never seenARC-AGI-3+19.8points ahead
Where Gemini 3.8 Flash pulls ahead
- Follows detailed instructions exactlyLiveBench · Instruction Following+17.6points ahead
- Short factual questions, answered correctlySimpleQA Verified+9.8points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+5.6points ahead
Every test, side by side
All 44 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 5
- LiveBench · Agentic CodingClaude Opus 5 by 1154.265.2+11
- LMArena · WebDevClaude Opus 5 by 104 rating points15861690+104 rating
- LiveBench · CodingClaude Opus 5 by 972.581.5+9
- Terminal-Bench 2.1Claude Opus 5 by 1.587.689.1+1.5
- SciCodetie56.656.4tie
AgentsClaude Opus 5
- Terminal-Bench 4.0Claude Opus 5 by 29.319.749+29.3
- GDPValClaude Opus 5 by 14.346.861.1+14.3
- τ-Bench V3 · BankingGemini 3.8 Flash by 2.844.942.1+2.8
ReasoningClaude Opus 5
- ARC-AGI-3Claude Opus 5 by 19.810.430.2+19.8
- CritPtClaude Opus 5 by 10.818.329.1+10.8
- Humanity's Last ExamClaude Opus 5 by 7.147.854.9+7.1
- GPQA DiamondGemini 3.8 Flash by 2.195.393.2+2.1
- LiveBench · ReasoningClaude Opus 5 by 1.989.391.2+1.9
- SimpleBenchGemini 3.8 Flash by 1.882.480.6+1.8
- ARC-AGI-2Claude Opus 5 by 1.289.290.4+1.2
FactsGemini 3.8 Flash
- SimpleQA VerifiedGemini 3.8 Flash by 9.869.759.9+9.8
- AA-Omniscience · AccuracyClaude Opus 5 by 6.354.660.9+6.3
- AA-Omniscience · Non-hallucinationGemini 3.8 Flash by 5.644.839.2+5.6
Images and chartsGemini 3.8 Flash
- CharXiv (RQ)Gemini 3.8 Flash by 2.586.283.7+2.5
- MMMU-Protie85.684.7tie
- LMArena · Visiontie13141319tie
MathClaude Opus 5
- FrontierMath Tier 4Claude Opus 5 by 51.221.973.2+51.2
- FrontierMath Tiers 1-3 (v2)Claude Opus 5 by 17.268.485.6+17.2
- LiveBench · MathematicsClaude Opus 5 by 4.191.695.7+4.1
Following instructionsGemini 3.8 Flash
- LiveBench · Instruction FollowingGemini 3.8 Flash by 17.681.463.8+17.6
Other results18 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.
- GDP (Surge AI)Claude Opus 5 by 50.53585.5+50.5
- Terminal-Bench 4.0Claude Opus 5 by 32.719.151.8+32.7
- GDPval-AA v2Claude Opus 5 by 279 rating points15451824+279 rating
- GDPval-AA v2 EloClaude Opus 5 by 279 rating points15451824+279 rating
- livebench_data_analysisClaude Opus 5 by 20.65474.5+20.6
- OSWorld 2.0 (partial)Claude Opus 5 by 16.45975.4+16.4
- HiL-Bench (Tools-allowed)Claude Opus 5 by 15.541.557+15.5
- AA Agentic IndexClaude Opus 5 by 15.141.156.2+15.1
- LVBenchGemini 3.8 Flash by 11.787.175.4+11.7
- OSWorld 2.0Claude Opus 5 by 11.65970.6+11.6
- AA IntelligenceClaude Opus 5 by 9.940.950.8+9.9
- AA-OmniscienceClaude Opus 5 by 7.529.637.1+7.5
- DeepSWE 1.1Gemini 3.8 Flash by 4.973.768.8+4.9
- GDP.PDF (All pass rate)Claude Opus 5 by 23537+2
- Artificial Analysis Coding IndexClaude Opus 5 by 1.776.378+1.7
- ARC-AGI-1Gemini 3.8 Flash by 198.597.5+1
- livebench_languagetie87.888.7tie
- HLE-Verifiedtie54.954.4tie
Questions people ask
Which is better, Gemini 3.8 Flash or Claude Opus 5?
Gemini 3.8 Flash and Claude Opus 5 each win four of the eight areas we test. Gemini 3.8 Flash wins facts, images and charts, long documents and following instructions; Claude Opus 5 wins coding, agents, reasoning and math. Gemini 3.8 Flash costs 85% less.
Which is better for coding?
Claude Opus 5. It wins 4 of the 5 coding tests both models report; Gemini 3.8 Flash wins none, and 1 is a tie.
Which is cheaper?
Gemini 3.8 Flash costs $0.75 per million input tokens and $3.75 per million output tokens; Claude Opus 5 costs $5.00 and $25.00. That makes Gemini 3.8 Flash about 85% cheaper for the same work.
How do you compare the two?
We use the 44 benchmark tests both models have published scores on. The verdict counts the 26 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 18 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.