Claude 3.5 Sonnet vs Claude 3 Sonnet
Wins 3 of 3 areas
Coding · Reasoning · Images and charts
Wins 0 of 3 areas
—
Claude 3.5 Sonnet is the stronger all-rounder.
Scores updated · 35 tests both models report · How we compare
Where each one wins
Tests won in each of the three 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.
Coding and images and charts 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 tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Claude 3.5 Sonnet pulls ahead
- Math problems shown in pictures and chartsMathVista+19.8points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+16points ahead
- Code for real scientific research problemsSciCode+8.7points ahead
Where Claude 3 Sonnet pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 35 tests both models report. The winning score is in its model's colour; marks a score checked independently.
ReasoningClaude 3.5 Sonnet
- GPQA DiamondClaude 3.5 Sonnet by 165640+16
- Humanity's Last Examtie3.23.6tie
Images and chartsClaude 3.5 Sonnet
- MathVistaClaude 3.5 Sonnet by 19.867.747.9+19.8
Other results31 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.
- NarrativeQAClaude 3.5 Sonnet by 63.574.611.1+63.5
- naturalquestions_closedbookClaude 3.5 Sonnet by 47.450.22.8+47.4
- MATHClaude 3.5 Sonnet by 2871.143.1+28
- GPQA (Diamond) Maj@32 5-shot CoTClaude 3.5 Sonnet by 20.967.246.3+20.9
- MMLU-ProClaude 3.5 Sonnet by 20.877.656.8+20.8
- HumanEvalClaude 3.5 Sonnet by 20.793.773+20.7
- GPQA (Diamond) 0-shot CoTClaude 3.5 Sonnet by 1959.440.4+19
- HumanEval 0-shotClaude 3.5 Sonnet by 199273+19
- MMMU (validation)Claude 3.5 Sonnet by 15.268.353.1+15.2
- BBQ Disambig Accuracy (%)Claude 3 Sonnet by 14.276.290.4+14.2
- MMLU 0-shot CoTClaude 3.5 Sonnet by 11.288.377.1+11.2
- MMLUClaude 3.5 Sonnet by 10.488.778.3+10.4
- BBHClaude 3.5 Sonnet by 10.293.182.9+10.2
- BIG-Bench Hard 3-shot CoTClaude 3.5 Sonnet by 10.293.182.9+10.2
- XSTestClaude 3.5 Sonnet by 9.895.685.8+9.8
- MMLU 5-shot CoTClaude 3.5 Sonnet by 8.990.481.5+8.9
- DROPClaude 3.5 Sonnet by 8.287.178.9+8.2
- MGSMClaude 3.5 Sonnet by 8.191.683.5+8.1
- MGSM 0-shot CoTClaude 3.5 Sonnet by 8.191.683.5+8.1
- DocVQA (test, ANLS score)Claude 3.5 Sonnet by 5.795.289.5+5.7
- OpenBookQAClaude 3.5 Sonnet by 5.497.291.8+5.4
- bbqClaude 3.5 Sonnet by 4.994.990+4.9
- BBQ Disambig Bias (%)Claude 3 Sonnet by 4.9-3.71.2+4.9
- BBQ Ambig Bias (%)Claude 3 Sonnet by 4.10.95+4.1
- GSM8KClaude 3.5 Sonnet by 4.196.492.3+4.1
- HarmBenchClaude 3.5 Sonnet by 2.398.195.8+2.3
- AA IntelligenceClaude 3.5 Sonnet by 1.37.25.9+1.3
- AIR-Bench 2024Claude 3.5 Sonnet by 1.285.984.7+1.2
- anthropic_red_teamtie99.899.8tie
- BBQ Ambig Accuracy (%)tie93.693.6tie
- simple_safety_teststie100100tie
Questions people ask
Which is better, Claude 3.5 Sonnet or Claude 3 Sonnet?
Claude 3.5 Sonnet wins all three areas where both have results: coding, reasoning and images and charts. Claude 3 Sonnet wins none.
Which is better for coding?
Claude 3.5 Sonnet. It wins the one coding test both models report.
How do you compare the two?
We use the 35 benchmark tests both models have published scores on. The verdict counts the 4 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 31 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.