Anthropic launches Claude Sonnet 5 with better coding, reasoning, and agentic search

Anthropic has launched one more AI model called Sonnet 5 with improved agentic capabilities along with stronger coding performance. It is available with lower operating costs as compared to previous Sonnet models. According to Anthropic, Sonnet 5 delivers better autonomous task handling. In addition, it also offers some capabilities that are close to its more powerful Opus 4.8 model.

Besides this launch, the AI company has also confirmed that that the US government has lifted export restrictions on its Fable 5 and Claude Mythos 5. This will allow broader deployment of Fable 5 globally.

Claude Sonnet 5 is launched with enhanced agentic performance

Sonnet 5 by Claude is now available worldwide. Anthropic has introduced an introductory price of $2 per million input tokens and $10 per million output tokens, until Aug. 31. Following this, the base fees will be $15 per million output tokens and $3 per million input tokens.

Anthropic has also raised rate limits for Chat, Cowork, Claude Code, and Claude Platform to accommodate the increased usage of tokens by the new model.

Much improved programming and logic problem-solving skills compared to Sonnet 4.6

According to Anthropic, Claude Sonnet 5 is significantly better than Sonnet 4.6, particularly when it comes to agentic coding.

In complex reasoning problems, Sonnet 5 is closer to Opus 4.8, but has a lower cost per task, Anthropic says.

Agentic search performance

Claude Sonnet 5’s agentic search capabilities are almost on par with Opus 4.8 on the BrowseComp benchmark test, and substantially outperforms Sonnet 4.6, Anthropic claims.

The company states that preliminary tests showed the model to be significantly improved in autonomous workflows. It could also be used to confirm its own answers, without having to be prompted, which made it more reliable for multi-step tasks.

Claude Sonnet 5 also enhances safety

In addition to performance, the model has some safety enhancements, says Anthropic.

Claude Sonnet 5 is better at refusing malicious requests and is more resistant to prompt injection attacks. The company also claims that it generates less hallucination and less sycophancy than Sonnet 4.6, which results in more reliable responses.

Fable 5 returns to global users with Anthropic’s help

In another development, Anthropic announced that the U.S. government has lifted its export restrictions on the Fable 5 AI model.

After the decision, Fable 5 has already been released world-wide via Claude Platform, Claude.ai, Claude Code, and Claude Cowork. Users with Anthropic plans other than Free will also see weekly usage raised by 50 percent until July 7.

Support for Fable 5 on AWS, Google Cloud and Microsoft Foundry is ongoing and will be rolled out as it progresses, the company said.

Access to Mythos 5 is still restricted

.Fable 5 is back for all users globally, but Claude Mythos 5 is back for users in a handful of US organisations only after permission was granted to Claude in the United States by the US government on 26th June.

Add Techlusive as a Preferred Source

Anthropic is looking to provide better AI performance, autonomous capabilities, and security with Claude Sonnet 5 that will not be as expensive as its flagship models. Concurrently, Fable 5 is an essential milestone in Fable’s efforts to build access to its more sophisticated AI capabilities.