Claude Sonnet 5

4.75
High‑performance agentic model for coding, tool use, and reasoning—secure, cost‑effective, and available across all plans.
Advertisement 728 × 90
CompanyAnthropic
Context1M
Released2026-06
Updated2026-07-17

Claude Sonnet 5 Overview

Claude Sonnet 5 is Anthropic’s latest agentic model—and it’s being called their most agent-capable Sonnet yet. It can autonomously plan and execute multi‑step, complex tasks, wield tools like browsers and terminals with ease, and handle end‑to‑end programming and knowledge work. It actively debugs on its own, without pausing midway. Compared to its predecessor, Sonnet 4.6, it shows major gains in reasoning, tool use, and coding—delivering performance that edges close to the more powerful Opus 4.8, but at a lower cost. It also offers stronger safety features, with network security protections enabled by default. The model is available across all Claude plans—free and Pro users get it as the default model.

Claude Sonnet 5 Pricing

PlanPriceDescription
API Usage Pricing Input $3 / Output $15 per 1M Tokens Supports large-scale API usage with long-context capabilities. Cache reads are available at a lower rate, making it suitable for developer integrations and enterprise applications.
API Cache Reads $0.30 per 1M Tokens Reduces API costs by optimizing repeated input processing through cached content.
Free Plan Free Provides limited access for testing and everyday use.
Pro Plan $20/month Offers higher usage limits and full Claude access, suitable for creators, developers, and professional users.
Max / Team / Enterprise Plans $100+/month / Custom Pricing Max provides higher usage limits; Team supports collaboration features; Enterprise offers advanced security and customized solutions.

Claude Sonnet 5 Key Features

①Powerful Agentic Capabilities
Autonomously plans and executes multi‑step, complex tasks, and can run independently for extended periods without stopping midway. Handles end‑to‑end workflows—such as complete software engineering projects—with minimal need for human intervention.

② Tool Use
Proficiently uses external tools like browsers and terminals. Performs strongly on benchmarks for agentic search and computer use.

③Coding & Debugging
Delivers excellent coding performance, capable of tackling complex software engineering tasks. Actively debugs on its own, even without being prompted to do so.

④Multi‑step Reasoning & Knowledge Work
Shows substantial improvements over the previous Sonnet 4.6 in reasoning, tool use, and coding. Approaches the performance of the more powerful Opus 4.8, while maintaining a lower cost.

⑤Enhanced Safety
Exhibits lower rates of undesirable behavior than Sonnet 4.6, with better safety performance in agentic scenarios. Network‑security capabilities are significantly more restricted than those of the Opus series, with network protection enabled by default.

Summary

①Developers & Software Engineering Teams
The core audience. Ideal for coding, debugging, and handling complex software engineering workflows—can autonomously complete end‑to‑end coding tasks.

②Teams & Enterprises Building AI Agents
Designed for those developing products or workflows that require autonomous planning, tool use (browsers, terminals, etc.), and long‑running independent operation.

③Knowledge Workers & Researchers
Well‑suited for those who need multi‑step reasoning, information retrieval (agentic search), and data organization and analysis.

④Cost‑Sensitive Users & Enterprises
Delivers performance close to Opus 4.8 at a lower price point—an excellent choice for teams seeking high cost‑effectiveness and large‑scale API usage.

⑤Security‑Conscious Scenarios
Features lower rates of undesirable behavior and default‑enabled network protection—ideal for enterprise‑level deployments with strict compliance requirements.

⑥Claude Users at All Tiers
Available across all plans: free and Pro users get it as the default, while Max, Team, and Enterprise plans can also access it—covering a broad range of users from individuals to large organizations.

Comments (0)

Leave a comment

Advertisement 728 × 90

Anthropic Model Comparison

Model Context Pricing API Released Global Heat
Claude Sonnet 5
1M Paid YES 2026-06
95/100
1M YES 2026-07
100/100
1M Paid YES 2026-06
93/100
1M Paid YES 2026-05
98/100
1M Paid YES 2026-05
96/100

Similar Models

Related Tools

Related News