Anthropic launches Claude Sonnet 5 AI model with coding, safety upgrades as Fable and Mythos controls lifted

2026-06-30 · Source: AI – SiliconANGLE · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy · Depth: Intermediate, quick

Summary

Anthropic PBC today debuted Claude Sonnet 5, a midrange large language model that outperforms its predecessor in several areas and will serve as the default option in the consumer tiers of the company's Claude chatbot service. The model improved its coding capabilities on SWE-Bench Pro and Terminal-Bench 2.1 benchmarks by 5.1% and 13.4% respectively, though OpenAI's GPT-5.6 Terra slightly outperformed it on Terminal-Bench 2.1 by about 4%. Sonnet 5 also scored 1,618 on GDPval-AA v2, up from Sonnet 4.5's 1,395 points, attributed to increased autonomy and improved safety features against malicious requests and prompt injection attacks. Additionally, Anthropic announced that U.S. government export controls on its flagship Mythos 5 and scaled-down Fable 5 models have been lifted, allowing their rollout to resume, with Fable 5 becoming broadly available and Mythos 5 accessible to trusted organizations. Sonnet 5 will be priced at \$3 per million input tokens and \$15 per million output tokens starting in September.

Key takeaway

For AI Engineers evaluating LLMs for coding and general knowledge tasks, Anthropic's Claude Sonnet 5 offers significant performance and safety upgrades over its predecessor. You should consider its improved autonomy and prompt injection defenses, especially when deploying it as a default chatbot option or via API, noting its competitive pricing starting in September. Additionally, prepare for the broader availability of Fable 5 and limited access to Mythos 5 following lifted export controls.

Key insights

Anthropic's new Sonnet 5 model enhances coding, autonomy, and safety, while Mythos 5 and Fable 5 controls are lifted.

Principles

• LLM autonomy can improve output quality.
• Increased autonomy requires enhanced safety guardrails.
• Midrange models are competitive in specific benchmarks.

Method

LLM capabilities are evaluated using benchmarks like SWE-Bench Pro, Terminal-Bench 2.1, and GDPval-AA v2 to measure coding and knowledge work performance.

In practice

• Use Sonnet 5 for improved coding tasks.
• Deploy Sonnet 5 as Claude chatbot default.
• Access Sonnet 5 via Anthropic's API.

Topics

• Claude Sonnet 5
• Large Language Models
• AI Safety
• Prompt Injection
• LLM Benchmarking
• Anthropic API

Best for: Machine Learning Engineer, NLP Engineer, CTO, AI Engineer, Director of AI/ML, Tech Journalist

Related on AIssential

• Anthropic's new Claude Sonnet 5 closes the gap to the pricier Opus model series — Sonnet 5 significantly boosts agentic capabilities and performance, closing the gap to Opus models while maintaining low cybersecurity risk.
• Anthropic's new Claude Fable 5 is the same base model as Mythos but with guardrails attached — Fable 5 offers Mythos-class AI power with integrated safety guardrails for broader, secure general use.
• Anthropic Claude Fable 5 — Anthropic's new Claude Fable 5 offers superior agentic capabilities but introduces controversial, silent performance limitations for specific research tasks.
• 😺 Fable 5 is back baby — Anthropic is democratizing agentic AI with Sonnet 5 while its frontier model, Fable 5, returns with usage caveats.
• Anthropic finally released Claude Fable 5, a public Mythos-class model. — Anthropic's Claude Fable 5 introduces a guarded public release model, balancing advanced capabilities with controlled access for sensitive AI research.
• Anthropic launches Claude Sonnet 5 as a cheaper way to run agents — Agentic capability is now a baseline expectation for foundation models, with cost and reliability becoming key differentiators.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI – SiliconANGLE.