Breaking

Anthropic Releases Claude Sonnet 5, Sets It as Default Across Plans

July 1, 2026 at 12:15 EDT

Foundation Models
AI Agents
Software Dev & Coding

Anthropic released Claude Sonnet 5, the latest model in its Sonnet family, on June 30, 2026, and made it the default model on the Free and Pro plans on day one. A successor to Sonnet 4.6, the company describes it in its announcement as "our most agentic Sonnet yet." According to the official announcement, it improves substantially over Sonnet 4.6 in reasoning, tool use, coding and knowledge work, aiming to deliver performance close to the higher-tier Opus 4.8 at Sonnet pricing.

June 30, 2026 · Anthropic

Claude Sonnet 5: the "most agentic Sonnet" chases Opus at Sonnet prices

A new default model built to run multi-step software work on its own — using browsers, terminals and tools — aiming to close the gap with the pricier Opus class while staying in the affordable tier.

token context window (max 128K output)

63.2%

agentic coding score (up from 58.1%)

1.0–1.35×

more tokens per text (new tokenizer trade-off)

Agentic coding — score to scale

Sonnet 5 narrows the gap toward Opus 4.8, well above Sonnet 4.6 (SWE-bench Pro comparable, %).

58.1%

Sonnet 4.6

63.2%

Sonnet 5

69.2%

Opus 4.8

Priced under Opus — per million tokens

Input / output cost, positioned as the cost-performance pick.

Sonnet 5 (intro → 8/31)

$2 / $10

then $3 / $15

Opus 4.8

$5 / $25

higher-end tier

Praise

Completes complex, multi-step tasks
Verifies and debugs its own output
Pushes past steps where earlier Sonnet stalled

Criticism

Can cost more yet trail Opus 4.8 in some cases
More hallucination and higher session usage
Unnatural or lecturing conversational tone

Continue reading

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