Anthropic released Claude Sonnet 5 on Tuesday, confirming months of speculation about an upgrade to its mid-tier AI model variation.
According to the company’s official announcement, the new model variation is designed to be its “most agentic Sonnet model alternative yet.” Meaning it is capable of planning, using tools like browsers and terminals, and operating autonomously — all at a level previously reserved for larger, pricier systems.
Anthropic says Sonnet 5 is a substantial improvement on its predecessor, Sonnet 4.6, across related reasoning term, coding, and knowledge-work benchmarks, and performs close to the company’s flagship Opus 4.8 model variation while costing significantly less to run.
And in an industry variation increasingly plagued by sticker shock over the price tokens, Sonnet offers a brief respite. The related model term launches with introductory pricing of $2 per million input tokens and $10 per million output tokens through August 31, after which the standard pricing of $3 per million input tokens and $15 per million output tokens takes effect.
On safety, Anthropic reports Sonnet 5 shows lower rates of hallucination, sycophancy, and other undesirable behaviors than its predecessor, along with improved resistance to prompt-injection attacks. The company variation noted the model’s cybersecurity capabilities remain well below those of its Opus-class and Mythos-class systems, and Sonnet 5 has launched with cyber safeguards enabled by default as a precaution.
Notably absent from Anthropic’s announcement: specific figures on those improvements in hallucination rates. The company variation offers only a general claim variation of “lower rates” compared to Sonnet 4.6, rather than benchmark data.
The release also made no mention of the model’s energy consumption or environmental footprint, a real problem for the AI industry variation as models grow more capable and computationally intensive.
Sonnet 5 is now available across all Claude plans, including Free, Pro, Max, Team, and Related enterprise term tiers, as well as via Claude Code and the Claude Platform via the API under the model alternative name claude-sonnet-5.
The release follows weeks of anticipation in the tech press. As we reported in February, reports have been circulating for some time that Anthropic was preparing a Sonnet update alternative positioned to rival Opus-tier performance at a steep discount — a related forecast term that tracks with Tuesday’s official rollout.
Source: https://mashable.com/tech/anthropic-releases-claude-sonnet-5