Model
Model explorer

Claude Sonnet 4.5

CLOSED
Anthropic · Claude 4 family · released Sep 29, 2025

Best coding/agent Sonnet at release, reported to sustain focus on complex tasks for 30+ hours.

ReasoningCodingVisionFunction callingTool useAgentic
1976.7
Elo · rank #59
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
AIMEMath87.0%#58
best: GPT-5.2 · 100.0%
ARC-AGI-2Reasoning13.6%#23
best: GPT-5.6 · 92.5%
BrowseCompAgents43.9%#35
best: Kimi K3 · 91.2%
CharXivVision67.0%#25
best: Muse Spark 1.1 (xhigh) · 88.4%
GAIAAgents74.5%#2
best: MiniMax-M2 · 75.7%
GDPval-AAAgents1276#24
best: Claude Fable 5 · 1932
GPQA DiamondReasoning83.4%#64
best: GPT-5.6 · 94.6%
Humanity's Last ExamReasoning33.6%#40
best: Claude Sonnet 5 · 57.4%
LiveCodeBenchCoding73.0%#61
best: DeepSeek-V4-Pro (Think Max) · 93.5%
MMLUKnowledge89.5%#14
best: OpenAI o3 · 92.9%
MMMU-ProVision67.5%#33
best: Claude Opus 4.7 · 85.5%
MMMUVision77.8%#28
best: Claude Fable 5 · 89.3%
OSWorld-VerifiedAgents61.4%#20
best: Claude Fable 5 · 85.0%
SWE-bench VerifiedCoding77.2%#31
best: Claude Fable 5 · 95.0%
τ²-Bench RetailAgents86.2%#4
best: Claude Sonnet 4.6 · 91.7%
τ²-Bench TelecomAgents98.0%#7
best: Claude Opus 4.6 · 99.3%
Terminal-Bench 2.0Coding51.0%#46
best: GPT-5.6 · 88.8%
Run it locally
Closed weights — available via API only. No local deployment.
Input / M tok
$3
Output / M tok
$15
Claude 4 family
Elo progression across releases
Claude Opus 4May 20251702.9Claude Sonnet 4May 20251594.8Claude Opus 4.1Aug 20251760.6Claude Sonnet 4.5Sep 20251976.7Claude Haiku 4.5Oct 20251590.8Claude Opus 4.5Nov 20252274.5Claude Opus 4.5 (High)Nov 20252301.0Claude Opus 4.5 (Medium)Nov 20252283.3Claude Opus 4.6Feb 20262459.4Claude Sonnet 4.6Feb 20262437.8Claude Opus 4.7Apr 20262702.0Claude Opus 4.8May 20262794.3
Resources
DOCSplatform.claude.comAPIAvailable via Anthropic API
API price $3/$15 · each benchmark row carries its own source badge (see methodology)