Breaking

Anthropic Ships Claude Sonnet 5 With Stronger Agent Skills at Lower Prices

July 1, 2026 at 05:14 EDT

Foundation Models
AI Agents
Software Dev & Coding

Anthropic on June 30, 2026 formally unveiled Claude Sonnet 5, the latest version of its mid-tier model, and made it generally available the same day. The release emphasizes sharply improved autonomous, agentic task execution while pricing the model below the higher-end Opus class.

June 30, 2026 · Anthropic

Claude Sonnet 5: near-Opus performance at Sonnet pricing

Anthropic's "most agentic Sonnet yet" ships same-day as a cheaper way to run agents — planning, tool use, and autonomous execution once reserved for larger models, now the default for Free and Pro plans.

token context window (inherited)

Same day

GA across web, API, Claude Code & Copilot

full exploit development success rate

Output price per 1M tokens

Sonnet 5 undercuts the higher-end Opus 4.8 — priced to run agents cheaply.

$10

Sonnet 5

intro rate

$15

Sonnet 5

after Aug 31

$25

Opus 4.8

flagship

Input tokens follow the same pattern: $2 / $3 (Sonnet 5) vs $5 (Opus 4.8) per 1M.

Praise

Strong coding & CLI task results
Better comment accuracy, deeper thinking
Improved follow-through & self-checking

Criticism

Some say "Opus 4.8 is cheaper & better"
More hallucination, higher session usage
Feels like a minor, not major, update

The bet: a cost-efficient execution layer for the agentic-capability race with OpenAI and Google. Free/Pro users gain by default — but many developers still see too little differentiation from Opus at this early stage.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.