Claude Opus 5.5 vs Gemini 4 Argon
Wins 5 of 7 areas
Coding · Agents · Reasoning · Math · Long documents
Wins 0 of 7 areas
—
Claude Opus 5.5 is the stronger all-rounder.Gemini 4 Argon is cheaper.
Both rank among the ten best models we track in facts.
Scores updated · 55 tests both models report · How we compare
Where each one wins
Tests won in each of the seven areas where both have results. 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 software20Claude Opus 5.52 of 2 tests
- AgentsCarrying out multi-step tasks on its own20Claude Opus 5.52 of 2 tests
- MathCompetition and research-level math20Claude Opus 5.52 of 2 tests
- ReasoningHard problems that need careful thinking21Claude Opus 5.52 of 4 tests · 1 tie
- Long documentsFinding answers in very long texts10Claude Opus 5.51 of 1 test
- FactsGetting facts right instead of making them up22Even2 each
- Images and chartsUnderstanding pictures, charts and video00Even0 each · 1 tie
Images and charts and long documents 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 4 Argon costs 50% 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 5.5 pulls ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+16.9points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+16.3points ahead
- Real work tasks from 44 professionsGDPVal+12.1points ahead
Where Gemini 4 Argon pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+43.5points ahead
- Short factual questions, answered correctlySimpleQA Verified+3.8points ahead
- Common-sense trick questionsSimpleBench+3.3points ahead
Every test, side by side
All 55 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 5.5
- LMArena · WebDevClaude Opus 5.5 by 137 rating points18161679+137 rating
- SciCodeClaude Opus 5.5 by 5.166.961.8+5.1
AgentsClaude Opus 5.5
- GDPValClaude Opus 5.5 by 12.168.356.2+12.1
- Terminal-Bench 4.0Claude Opus 5.5 by 2.559.657.1+2.5
ReasoningClaude Opus 5.5
- CritPtClaude Opus 5.5 by 4.631.727.1+4.6
- Humanity's Last ExamClaude Opus 5.5 by 4.361.457.1+4.3
- SimpleBenchGemini 4 Argon by 3.388.491.7+3.3
- ARC-AGI-2tie91.791.3tie
FactsEven
- AA-Omniscience · Non-hallucinationGemini 4 Argon by 43.541.484.9+43.5
- AA-Omniscience · AccuracyClaude Opus 5.5 by 16.366.249.9+16.3
- SimpleQA VerifiedGemini 4 Argon by 3.872.276+3.8
- Vectara HHEM hallucination ratelower is betterClaude Opus 5.5 by 2.29.912.1+2.2
MathClaude Opus 5.5
- FrontierMath Tier 4Claude Opus 5.5 by 16.99578+16.9
- FrontierMath Tiers 1-3 (v2)Claude Opus 5.5 by 5.291.286+5.2
Other results39 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.
- GraphWalks (256k to 1M, BFS (F1))Gemini 4 Argon by 17.466.884.2+17.4
- CritPt v2 HardClaude Opus 5.5 by 15.56852.5+15.5
- OSWorld 2.0Claude Opus 5.5 by 12.681.869.2+12.6
- Computer use (OSWorld-2.1)Claude Opus 5.5 by 12.587.274.7+12.5
- HLE-DiamondClaude Opus 5.5 by 11.267.956.7+11.2
- GraphWalks (Up to 128k, BFS (F1))Gemini 4 Argon by 9.190.699.7+9.1
- Agentic coding (Terminal-bench 4.0)Claude Opus 5.5 by 966.457.4+9
- Terminal-Bench 4.0Claude Opus 5.5 by 966.457.4+9
- AutomationBench (Score)Gemini 4 Argon by 8.842.551.3+8.8
- Knowledge work (AutomationBench)Gemini 4 Argon by 8.842.551.3+8.8
- General capabilities (AA Index v5.0)Claude Opus 5.5 by 85850+8
- LVBenchGemini 4 Argon by 883.791.7+8
- Multimodal understanding (LVBench)Gemini 4 Argon by 883.791.7+8
- HealthBenchClaude Opus 5.5 by 7.95749.1+7.9
- RiemannBenchGemini 4 Argon by 6.469.676+6.4
- Science and math (RiemannBench)Gemini 4 Argon by 6.469.676+6.4
- Science and math (Terminal-Bench-Science 0.1 (AA))Claude Opus 5.5 by 6.15952.9+6.1
- Terminal-Bench-Science 0.1Claude Opus 5.5 by 6.15952.9+6.1
- Science and math (Terminal-Bench Science 0.1)Claude Opus 5.5 by 5.763.357.6+5.7
- AA IntelligenceClaude Opus 5.5 by 557.652.6+5
- HealthBench ProfessionalClaude Opus 5.5 by 4.859.955.1+4.8
- GDP (Surge AI)Claude Opus 5.5 by 4.426.221.8+4.4
- AA-OmniscienceClaude Opus 5.5 by 446.442.4+4
- PostTrainBenchClaude Opus 5.5 by 449.345.3+4
- DeepSWE 1.1Gemini 4 Argon by 3.774.277.9+3.7
- HiL-Bench (Tools-allowed)Gemini 4 Argon by 35962+3
- Computer use (Agents' Last Exam v1.1)Claude Opus 5.5 by 2.671.669+2.6
- Agentic coding (Terminal-Bench 4.0 (AA))Claude Opus 5.5 by 2.559.657.1+2.5
- vectara_factual_consistencyClaude Opus 5.5 by 2.290.187.9+2.2
- Knowledge work (Vals Index)Gemini 4 Argon by 1.96768.9+1.9
- Vals IndexGemini 4 Argon by 1.96768.9+1.9
- vectara_avg_summary_lengthClaude Opus 5.5 by 13.9 rating points192.1178.2+13.9 rating
- Agent's Last Exam (Pass rate)Gemini 4 Argon by 1.338.239.5+1.3
- Computer use (Agent's Last Exam)Gemini 4 Argon by 1.338.239.5+1.3
- ARC-AGI-1Gemini 4 Argon by 197.598.5+1
- Cybersecurity (CWE-bench v1)Gemini 4 Argon by 16768+1
- Cybersecurity (CWE-bench)Gemini 4 Argon by 16768+1
- General capabilities (Vals Index v2.2)Claude Opus 5.5 by 170.269.2+1
- vectara_answer_ratetie99.799.1tie
Questions people ask
Which is better, Claude Opus 5.5 or Gemini 4 Argon?
Claude Opus 5.5 wins five of the seven areas where both have results: coding, agents, reasoning, math and long documents. Gemini 4 Argon wins none, but costs 50% less. They are level on facts and images and charts.
Which is better for coding?
Claude Opus 5.5. It wins 2 of the 2 coding tests both models report; Gemini 4 Argon wins none.
Which is cheaper?
Claude Opus 5.5 costs $4.00 per million input tokens and $20.00 per million output tokens; Gemini 4 Argon costs $2.00 and $10.00. That makes Gemini 4 Argon about 50% cheaper for the same work.
How do you compare the two?
We use the 55 benchmark tests both models have published scores on. The verdict counts the 16 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 39 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.