Model
Model explorer

Claude Sonnet 4.6

CLOSED
Anthropic · Claude 4 family · released Feb 17, 2026

CONFIRMED REAL — this is not a mix-up. Model id claude-sonnet-4-6. Direct predecessor to Claude Sonnet 5, sitting between Sonnet 4.5 and Sonnet 5 in the lineage; the corpus is missing it even though it already has both Opus 4.6 (2026-02-05) and Sonnet 5 (2026-06-30). Introduced 1M-token context (beta) and adaptive thinking for the Sonnet line.

ReasoningCodingVisionFunction callingTool useAgentic
2370.1
Elo · rank #53
Parameters
Undisclosed
Active params
—
Context
1M 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
Agents' Last ExamAgents32.0%#15
best: GPT-6 Astra · 59.3%
AIMEMath95.6%#12
best: GPT-5.2 · 100.0%
ARC-AGI-2Reasoning58.3%#24
best: GPT-6 Astra · 95.0%
Arena EloHuman preference1472#25
best: Claude Fable 5 · 1504
BrowseCompAgents74.0%#25
best: Kimi K3 · 91.2%
CharXivVision72.4%#29
best: Qwen3.8-Flash-Next · 90.6%
GDPval-AAAgents1633#11
best: Claude Fable 5 · 1932
GPQA DiamondReasoning89.9%#37
best: GPT-6 Astra · 96.0%
Humanity's Last ExamReasoning49.0%#14
best: Claude Opus 5.5 · 67.7%
LiveCodeBenchCoding82.1%#42
best: DeepSeek-V4-Pro (Think Max) · 93.5%
MMLU-ProKnowledge87.3%#10
best: Claude Fable 5 · 91.5%
MMMLUKnowledge89.3%#9
best: Gemini 3.1 Pro · 92.6%
MMMU-ProVision83.6%#3
best: Claude Opus 4.7 · 85.5%
OSWorld-VerifiedAgents72.5%#18
best: Claude Fable 5 · 85.0%
SWE-bench ProCoding58.1%#25
best: Claude Opus 5.5 · 89.9%
SWE-bench VerifiedCoding77.4%#41
best: Claude Opus 5 · 97.0%
τ²-Bench RetailAgents91.7%#1
best: this model · 91.7%
τ²-Bench TelecomAgents97.9%#10
best: Claude Opus 4.6 · 99.3%
Terminal-Bench 2.0Coding67.0%#36
best: DeepSeek-V4.1-Flash · 90.6%
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 20251637.6Claude Sonnet 4May 20251533.4Claude Opus 4.1Aug 20251700.2Claude Sonnet 4.5Sep 20251905.6Claude Haiku 4.5Oct 20251518.6Claude Opus 4.5Nov 20252187.9Claude Opus 4.5 (High)Nov 20252220.5Claude Opus 4.5 (Medium)Nov 20252266.8Claude Opus 4.6Feb 20262420.5Claude Sonnet 4.6Feb 20262370.1Claude Opus 4.7Apr 20262650.0Claude Opus 4.8May 20262670.2
Resources
DOCShttps://platform.claude.com/docs/en/about-claude/models/overviewAPIAvailable via Anthropic API
API price $3/$15 · each benchmark row carries its own source badge (see methodology)