Back to Model Tracker
Anthropic Active

Claude Sonnet 4.6

Anthropic’s best speed-to-intelligence balance, with strong reasoning and coding plus the full 1M-token context window.

Claude 4 Updated May 20, 2026
Post Reboot Practical Scores

Capability profile

90.1
Reasoning 91.0
Coding 92.0
Writing 92.0
Speed 90.0
Cost efficiency 74.0
Image understanding 88.0
Pricing
Input
$3 / MTok
Output
$15 / MTok
Cached input
$0.30 / MTok cache hit
Context
1M tokens
Free tier
No / unknown
API
Available
Subscription
Claude API, claude.ai, Bedrock, Vertex AI, Foundry
Open-source status
Closed
Best Use Cases
Best for
  • Balanced production use where you want strong reasoning
  • speed
  • and long context without paying Opus rates.
Best workflows
  • General agents
  • code assistance
  • enterprise search/synthesis
  • and document workflows.
Weakest at
  • Absolute frontier performance comparisons against the strongest premium models.
Not ideal for
  • Ultra-cheap inference or purely lightweight chatbot use.
Post Reboot Take

Sonnet is often the pragmatic Anthropic choice. It gives you enough intelligence to do real work without immediately making every request feel expensive.

Anthropic’s best speed-to-intelligence balance, with strong reasoning and coding plus the full 1M-token context window.

Post Reboot take

Sonnet is often the pragmatic Anthropic choice. It gives you enough intelligence to do real work without immediately making every request feel expensive.

Strengths

Best for: Balanced production use where you want strong reasoning, speed, and long context without paying Opus rates.

Weakest at: Absolute frontier performance comparisons against the strongest premium models.

Best workflows: General agents, code assistance, enterprise search/synthesis, and document workflows.

Example use cases: Knowledge workers, internal copilots, ops analysis, and product-side AI features.

Sources used for this profile