Model

Claude 3.7 Sonnet

CLOSED

Anthropic · Claude 3.7 family · released Feb 24, 2025

First hybrid-reasoning Claude with visible extended thinking; launched alongside Claude Code; retired Feb 2026.

ReasoningCodingVisionFunction callingTool useAgentic

1450.1

Elo · rank #131

Parameters

Undisclosed

Active params

—

Context

200K tokens

Architecture

Dense transformer, hybrid-reasoning (parameter count 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

Aider PolyglotCoding60.4%#18

best: Claude Opus 4.5 · 89.4%

AIMEMath23.3%#139

best: GPT-5.2 · 100.0%

Arena-Hard v2Human preference59.8%#6

best: Qwen3-Max · 86.1%

GAIAAgents56.4%#5

best: MiniMax-M2 · 75.7%

GPQA DiamondReasoning68.0%#131

best: GPT-5.6 · 94.6%

IFEvalReasoning90.8%#13

best: Gemma 4 26B A4B · 98.5%

LiveCodeBenchCoding56.7%#106

best: DeepSeek-V4-Pro (Think Max) · 93.5%

MATH-500Math82.2%#63

best: GPT-5 · 99.4%

MMLU-ProKnowledge80.7%#53

best: Claude Fable 5 · 91.5%

MMMLUKnowledge83.2%#21

best: Gemini 3.1 Pro · 92.6%

MMMUVision71.8%#48

best: Claude Fable 5 · 89.3%

PutnamBenchMath0.0%#7

best: DeepSeek-Prover-V2 671B · 7.1%

SWE-bench VerifiedCoding63.7%#71

best: Claude Fable 5 · 95.0%

tau-benchAgents81.2%#3

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 3.7 family

Elo progression across releases

Claude 3.7 SonnetFeb 20251450.1

Resources

DOCSplatform.claude.com

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