LANGUAGE MODEL Anthropic Last updated:

Claude Sonnet 4.6

The Production Workhorse Update

Anthropic's mid-cycle November 2025 Sonnet refresh — a capability bump rather than a major-version release. Sonnet 4.6 became the default coding model behind most AI IDEs in Q4 2025, displacing both Sonnet 4 and Opus 4 in volume usage (since Opus's premium price made it overkill for most interactive coding sessions).

Intelligence
Top 10%
Speed
Slow
63 tok/s output
Cost
High
$3.75 in / $15.00 out
Context
200K
Up to 200,000 tokens
How are Intelligence, Speed & Cost bucketed?
Intelligence and Speed buckets are percentile ranks on Artificial Analysis. Cost buckets are fixed dollar thresholds keyed off output-token price ($/M out).
Intelligence
  • Top 1%≤ 1%
  • Top 5%≤ 5%
  • Top 10%≤ 10%
  • Good≤ 25%
  • Medium≤ 50%
  • Below avg> 50%
Speed
  • Top 1%≥ 345 tok/s
  • Top 5%≥ 237 tok/s
  • Top 10%≥ 196 tok/s
  • Good≥ 146 tok/s
  • Medium≥ 90 tok/s
  • Slow< 90 tok/s
Cost
  • Freeopen weights · self-host
  • Low< $1 / M out
  • Moderate$1–5 / M out
  • High≥ $5 / M out

Why it matters

Sonnet 4.6 is what most production AI workflows actually run on as of early 2026 — including the IDE you're probably using to view this code. The model's specific identifier appears in this project's CLAUDE knowledge prompt as `claude-sonnet-4-6` — which is the lineage marker for the workhorse tier of the family.

Core Capabilities

Long Documents
Handles entire codebases, books, and multi-doc RAG.
Multimodal
Combines text, vision, and audio in one model.
Generative
Produces images, video, audio, or other media.
Agent Workflows
Built for tool use and autonomous tasks.

Context Window

200k tokens
≈ 154+ pages
4k Chat 聊天
32k Long docs 长文档
128k Books 整本书
400k Multi-doc 多文档
1M Codebase 整个代码库
10M
200k

Availability

API
Available
Product / App
Available
Open Source
Not released
Enterprise
Available

Pricing Model

Pay per token
Input and output billed separately.
Pay-per-token

Capability / Performance

Where this model sits relative to the middle 60% of models in the tree. All scores are 0–10 (higher is better).

Lower 20% Upper 80% This model
Reasoning
AA Intelligence Index · scaled to 10
1.7
5.6
7.4
Coding
SciCode · scaled to 10
1.8
4.3
4.7
Agentic tasks
Terminal-Bench Hard · scaled to 10
0.2
3.6
5.3
Context / memory
Context window size · log-scaled
6.0
9.0
6.7
Cost efficiency
Input price ($/M tokens) · cheaper scores higher
6.2
10.0
4.5
Lower 20% 20th percentile — 20% of models score below this This model Where the current model lands Upper 80% 80th percentile — only 20% of models score above this Percentile boundaries are computed across every model in the tree that reports the underlying benchmark for each capability.

What it feels like

  • Topped SWE-bench Verified at release — Cognition reported 18% planning gain and 12% end-to-end eval gain for Devin
  • Edit accuracy is the headline change — internal code-editing error rate dropped from 9% on Sonnet 4 to 0%
  • Sustains 30+ hour autonomous coding focus without losing the thread
  • Reasoning quality improvements are real but incremental over 4.5 — pick this for cost/speed, not for raw IQ
  • Same price point as predecessor — easy upgrade if you're already on Sonnet
  • Some testers find Opus 4.5 worth the premium for the hardest multi-system bugs

Best use cases

  • Production agentic coding (Devin, Cursor, Cline) where reliability matters
  • AI SRE / on-call automation with long task horizons
  • Day-to-day software engineering at scale on real GitHub issues
  • Cost-sensitive teams that need frontier-tier coding without Opus pricing

Tools to try

Not ideal for

  • The very hardest reasoning tasks — Opus 4.5 still leads by ~7 points on AA Intelligence Index
  • Simple chat or casual Q&A — try Haiku 4.5 instead
  • Workloads where 1M-token context is required (Sonnet 4.6 ships 200K)

Model Evolution

claude-sonnet is Anthropic's language model family.

View full evolution tree →