Claude Sonnet 4.6 Brings 1M Token Context Window to All Developers

1M Context Now Available at Sonnet Pricing

Anthropic has extended its 1 million token context window capability to Claude Sonnet 4.6, making massive context processing accessible at mid-tier pricing for the first time.

What 1M Tokens Means

• 750,000 words of text

• 3,000 pages of documents

• 150,000 lines of code

• 5-10 complete codebases

Context Compaction

New to Sonnet 4.6, context compaction automatically manages long conversations:

• When approaching the limit, older messages are summarized

• Critical information is preserved; verbose details are compressed

• Enables effectively unlimited conversation length

• Transparent to users—no manual management needed

Pricing Structure

Context Size

Input Price

Output Price

0-200K tokens

$3/M

$15/M

200K-1M tokens

$6/M

$30/M

Premium rates apply only to the portion exceeding 200K tokens.

Use Cases Unlocked

Full Codebase Analysis

"We can now load our entire monorepo—250K lines of TypeScript—and ask Claude to find all instances of a deprecated pattern," reports a senior engineer at a Series C startup.

Legal Document Review

Law firms are using 1M context to analyze complete contract sets, identifying conflicts and obligations across dozens of documents simultaneously.

Research Synthesis

Academic teams load multiple papers (10-15 full PDFs) and generate comprehensive literature reviews in single sessions.

Developer Response

GitHub discussions show immediate adoption:

"Finally can analyze our entire service mesh configuration in one prompt. Game changer for infrastructure audits."

"Loaded our complete API documentation—every endpoint, every type definition. Claude found 12 inconsistencies we'd missed for months."

Limitations

• Beta feature—expect occasional issues

• Quality degrades for needle-in-haystack at extreme lengths

• Opus 4.6 still superior for retrieval (76% vs ~18% on MRCR)

• Latency increases with context size

Getting Started

python
response = client.messages.create(
    model="claude-sonnet-4-6-20260217",
    max_tokens=8192,
    betas=["max-tokens-1m"],
    messages=[{"role": "user", "content": massive_document}]
)


What's Next

Anthropic hints at further context improvements in the Claude 5 family, with some analysts speculating 2M+ context windows by year end.