Skip to main content
Language modelActive

Claude Sonnet 5

Anthropic

Released
June 30, 2026
Data date
September 8, 2026
Benchmark-Meta-Score
52.3 / 100
Provisional, 59.3 % panel coverage
Speed
73.2 tokens/s
Output throughput
Price
$2 · $10
Input · Output / 1M tokens
Input
Text, Image
Output
Text

New in Claude Sonnet 5 was a tokenizer that Anthropic says can turn the same text into up to 35% more tokens, depending on its content. It followed Claude Sonnet 4.6 and has since been superseded by Claude Sonnet 5.5. In Anthropic’s BrowseComp and OSWorld-Verified comparisons, high-effort performance can match Claude Opus 4.8 on some tasks.

Over Sonnet 4.6, Anthropic reports fewer hallucinations and stronger prompt-injection resistance. The list price alone therefore does not capture that possible increase in token use.

Position
Rank 21 of 24
Index score
27.5 / 100
1.GPT-6.1 Sol90 / 100
2.GPT-6 Astra85.6 / 100
3.Claude Opus 578.8 / 100
4.GPT-6 Sol77.5 / 100
5.Grok 4.676.9 / 100
6.Grok 4.772.5 / 100
7.Gemini 3.8 Flash66.3 / 100
8.GPT-5.6 Sol65.6 / 100
9.GPT-5.561.9 / 100
10.Gemini 3.1 Pro Preview60.1 / 100
11.Grok 4.558.8 / 100
12.Gemini 3.7 Flash58.1 / 100
13.Gemini 3 Pro56.9 / 100
14.Claude Haiku 5.552.5 / 100
15.Gemini 3.6 Flash43.1 / 100
16.GPT-5.240.1 / 100
17.GPT-5.6 Luna33.1 / 100
17.GPT-6 Luna33.1 / 100
19.GPT-5.6 Terra31.3 / 100
20.Gemini 3.5 Flash28.8 / 100
21.Claude Sonnet 527.5 / 100
22.Grok 4.325.2 / 100
23.GPT-5 mini5 / 100
24.o3-mini1.3 / 100
Methodology and source

The LLM comparison hub's selected-panel index averages rank percentiles from 10 documented comparison cohorts. Benchmark meta-scores are shown in the benchmark section.