Model

Claude 3.5 Sonnet (New)

CLOSED

Anthropic · Claude 3.5 family · released Oct 22, 2024

Model id claude-3-5-sonnet-20241022, the corpus-confirmed gap: an upgraded revision of Claude 3.5 Sonnet released Oct 22 2024 (same price/speed as the June 2024 original) that introduced computer use (GUI automation: screenshots, cursor moves, clicks, typing) in public beta — the first frontier model to do so. Distinct from the June 2024 claude-3-5-sonnet-20240620 already in the corpus. Given the current corpus date frame (July 2026) this snapshot is now superseded/deprecated in Anthropic's lineup, though it was 'released' (not merely announced) at the time.

ReasoningCodingVisionFunction callingTool useAgentic

1357.3

Elo · rank #146

Parameters

Undisclosed

Active params

—

Context

200K tokens

Architecture

Dense transformer, architecture undisclosed

License

Proprietary

Languages

—

API price (in/out)

$3 / $15

Modalities

text · vision

Benchmark results

Bar shows position within the tracked field; marker = field best

AI2DVision95.3%#5

best: Molmo 72B · 96.3%

AIMEMath16.0%#144

best: GPT-5.2 · 100.0%

Arena EloHuman preference1373#28

best: Claude Fable 5 · 1505

BIG-Bench HardReasoning93.2%#2

best: ERNIE 4.5 300B-A47B · 94.3%

ChartQAVision90.8%#3

best: MiniMax-VL-01 · 91.7%

DocVQAVision94.2%#11

best: Qwen2-VL-72B · 96.5%

DROPReasoning88.3#10

best: Hunyuan-T1 · 93.1

FrontierMath (Tiers 1-3)Math2.1%#13

best: GPT-5.6 · 89.0%

GPQA DiamondReasoning60.0%#165

best: GPT-5.6 · 94.6%

Humanity's Last ExamReasoning4.0%#126

best: Claude Sonnet 5 · 57.4%

HumanEvalCoding93.7%#9

best: Claude Opus 4.5 · 99.4%

IFEvalReasoning90.2%#15

best: Gemma 4 26B A4B · 98.5%

MATH-500Math78.3%#70

best: GPT-5 · 99.4%

MathVistaVision70.7%#24

best: Seed 2.1 Pro · 90.7%

MGSMMath92.5%#2

best: OpenAI o4-mini · 93.7%

MMLU-ProKnowledge78.0%#64

best: Claude Fable 5 · 91.5%

MMLUKnowledge88.7%#19

best: OpenAI o3 · 92.9%

MMMUVision70.4%#51

best: Claude Fable 5 · 89.3%

OSWorldAgents14.9%#4

best: Seed 2.1 Pro · 78.8%

SWE-bench VerifiedCoding49.0%#92

best: Claude Fable 5 · 95.0%

tau-benchAgents69.2%#8

best: Claude Opus 4.1 · 82.4%

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

Output / M tok

$15

Claude 3.5 family

Elo progression across releases

Claude 3.5 SonnetJun 20241294.7 Claude 3.5 HaikuOct 20241041.8 Claude 3.5 Sonnet (New)Oct 20241357.3

Resources

DOCShttps://www.anthropic.com/news/3-5-models-and-computer-use APIAvailable via Anthropic API

API price $3/$15 · each benchmark row carries its own source badge (see methodology)