Compare · ModelsLive · 2 picked · head to head
Claude Opus 5 vs Gemini 3.8 Flash
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Claude Opus 5 wins on 13/18 benchmarks
Claude Opus 5 wins 13 of 18 shared benchmarks. Leads in agentic · coding · general.
Category leads
agentic·Claude Opus 5reasoning·Gemini 3.8 Flashknowledge·Gemini 3.8 Flashcoding·Claude Opus 5general·Claude Opus 5math·Claude Opus 5
Hype vs Reality
Attention vs performance
Claude Opus 5
#21 by perf·#9 by attention
Gemini 3.8 Flash
#53 by perf·#6 by attention
Best value
Gemini 3.8 Flash
5.7x better value than Claude Opus 5
Claude Opus 5
4.6 pts/$
$15.00/M
Gemini 3.8 Flash
26.5 pts/$
$2.25/M
Vendor risk
Who is behind the model
Anthropic
$965.0B·Tier 1
Google DeepMind
$4.27T·Tier 1
Head to head
18 benchmarks · 2 models
Claude Opus 5Gemini 3.8 Flash
APEX-Agents
Claude Opus 5 leads by +1.5
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
Claude Opus 5
65.8
Gemini 3.8 Flash
64.3
ARC-AGI
Gemini 3.8 Flash leads by +1.0
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
Claude Opus 5
97.5
Gemini 3.8 Flash
98.5
ARC-AGI-2
Claude Opus 5 leads by +1.3
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
Claude Opus 5
90.4
Gemini 3.8 Flash
89.2
Chess Puzzles
Gemini 3.8 Flash leads by +15.8
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Claude Opus 5
40.0
Gemini 3.8 Flash
55.8
Deepswe
Gemini 3.8 Flash leads by +0.2
Claude Opus 5
73.7
Gemini 3.8 Flash
73.8
Dtbench
Claude Opus 5 leads by +3.6
Claude Opus 5
96.5
Gemini 3.8 Flash
92.9
Frontiercode
Claude Opus 5 leads by +12.2
Claude Opus 5
53.4
Gemini 3.8 Flash
41.2
FrontierMath-Tier-4-v2-Private
Claude Opus 5 leads by +51.2
Claude Opus 5
73.2
Gemini 3.8 Flash
21.9
FrontierMath-Tiers-1-3-v2-Private
Claude Opus 5 leads by +17.2
Claude Opus 5
85.6
Gemini 3.8 Flash
68.4
Frontierswe
Claude Opus 5 leads by +32.4
Claude Opus 5
52.0
Gemini 3.8 Flash
19.6
Furniture Assembly
Claude Opus 5 leads by +41.7
Claude Opus 5
44.0
Gemini 3.8 Flash
2.4
GPQA diamond
Gemini 3.8 Flash leads by +2.0
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Claude Opus 5
91.8
Gemini 3.8 Flash
93.9
Lmca
Claude Opus 5 leads by +13.6
Claude Opus 5
75.8
Gemini 3.8 Flash
62.2
Mystery Game Puzzles
Claude Opus 5 leads by +13.2
Claude Opus 5
54.8
Gemini 3.8 Flash
41.6
OTIS Mock AIME 2024-2025
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Claude Opus 5
98.9
Gemini 3.8 Flash
98.9
Proofbench
Claude Opus 5 leads by +51.0
Claude Opus 5
99.0
Gemini 3.8 Flash
48.0
SimpleQA Verified
Gemini 3.8 Flash leads by +9.8
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
Claude Opus 5
59.9
Gemini 3.8 Flash
69.7
WeirdML
Claude Opus 5 leads by +7.0
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
Claude Opus 5
91.8
Gemini 3.8 Flash
84.8
Full benchmark table
| Benchmark | Claude Opus 5 | Gemini 3.8 Flash |
|---|---|---|
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | 65.8 | 64.3 |
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 97.5 | 98.5 |
ARC-AGI-2 ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data. | 90.4 | 89.2 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 40.0 | 55.8 |
Deepswe | 73.7 | 73.8 |
Dtbench | 96.5 | 92.9 |
Frontiercode | 53.4 | 41.2 |
FrontierMath-Tier-4-v2-Private | 73.2 | 21.9 |
FrontierMath-Tiers-1-3-v2-Private | 85.6 | 68.4 |
Frontierswe | 52.0 | 19.6 |
Furniture Assembly | 44.0 | 2.4 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 91.8 | 93.9 |
Lmca | 75.8 | 62.2 |
Mystery Game Puzzles | 54.8 | 41.6 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 98.9 | 98.9 |
Proofbench | 99.0 | 48.0 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 59.9 | 69.7 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 91.8 | 84.8 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $5.00 | $25.00 | 1.0M tokens (~500 books) | $100.00 | |
| $0.75 | $3.75 | 1.0M tokens (~524 books) | $15.00 |