Compare · ModelsLive · 2 picked · head to head
R1 vs Claude Sonnet 4.5
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Claude Sonnet 4.5 wins on 8/9 benchmarks
Claude Sonnet 4.5 wins 8 of 9 shared benchmarks. Leads in reasoning · knowledge · math.
Category leads
reasoning·Claude Sonnet 4.5knowledge·Claude Sonnet 4.5math·Claude Sonnet 4.5coding·Claude Sonnet 4.5
Hype vs Reality
Attention vs performance
R1
#116 by perf·no signal
Claude Sonnet 4.5
#132 by perf·no signal
Best value
R1
6.0x better value than Claude Sonnet 4.5
R1
28.2 pts/$
$1.60/M
Claude Sonnet 4.5
4.7 pts/$
$9.00/M
Vendor risk
Mixed exposure
One or more vendors flagged
DeepSeek
$3.4B·Tier 1
Anthropic
$380.0B·Tier 1
Head to head
9 benchmarks · 2 models
R1Claude Sonnet 4.5
ARC-AGI
Claude Sonnet 4.5 leads by +47.9
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
R1
15.8
Claude Sonnet 4.5
63.7
ARC-AGI-2
Claude Sonnet 4.5 leads by +12.3
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
R1
1.3
Claude Sonnet 4.5
13.6
DeepResearch Bench
Claude Sonnet 4.5 leads by +17.5
DeepResearch Bench · evaluates AI on complex multi-step research tasks requiring information gathering, synthesis, and producing comprehensive analyses.
R1
35.1
Claude Sonnet 4.5
52.6
GPQA diamond
Claude Sonnet 4.5 leads by +14.1
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
R1
62.3
Claude Sonnet 4.5
76.4
MATH level 5
Claude Sonnet 4.5 leads by +4.7
MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics.
R1
93.0
Claude Sonnet 4.5
97.7
OTIS Mock AIME 2024-2025
Claude Sonnet 4.5 leads by +24.5
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
R1
53.3
Claude Sonnet 4.5
77.8
SimpleBench
Claude Sonnet 4.5 leads by +28.1
SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking.
R1
17.1
Claude Sonnet 4.5
45.2
SimpleQA Verified
R1 leads by +3.8
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
R1
27.4
Claude Sonnet 4.5
23.6
WeirdML
Claude Sonnet 4.5 leads by +11.2
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
R1
36.5
Claude Sonnet 4.5
47.7
Full benchmark table
| Benchmark | R1 | Claude Sonnet 4.5 |
|---|---|---|
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 15.8 | 63.7 |
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. | 1.3 | 13.6 |
DeepResearch Bench DeepResearch Bench · evaluates AI on complex multi-step research tasks requiring information gathering, synthesis, and producing comprehensive analyses. | 35.1 | 52.6 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 62.3 | 76.4 |
MATH level 5 MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics. | 93.0 | 97.7 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 53.3 | 77.8 |
SimpleBench SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking. | 17.1 | 45.2 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 27.4 | 23.6 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 36.5 | 47.7 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.70 | $2.50 | 64K tokens (~32 books) | $11.50 | |
| $3.00 | $15.00 | 1.0M tokens (~500 books) | $60.00 |