Anthropic has launched Claude Sonnet 5, and the headline is not just that it is smarter, it is that it is dramatically cheaper to run as an autonomous agent. Released on June 30, 2026, Sonnet 5 becomes the default model for free and Pro users and lands with introductory pricing that undercuts most of its frontier rivals. For any business trying to put AI agents to work on real tasks, this launch changes the math.
The pricing is the story. Through August 31, 2026, Sonnet 5 costs $2 per million input tokens and $10 per million output tokens. After that promotional window, it settles at $3 per million input and $15 per million output. Even at standard pricing, that makes it cheaper than Anthropic’s own Opus 4.8, as well as GPT-5.5 and Gemini 3.1 Pro. It is more expensive than a lightweight option like Gemini 3.5 Flash, but the comparison that matters is capability per dollar, and on that axis Sonnet 5 is aggressive.
What you get for that price is a model built for agentic work. Anthropic describes Sonnet 5 as able to make plans, use tools like browsers and terminals, and run autonomously at levels that previously required larger, costlier models. On agentic coding benchmarks it scores 63.2%, up from Sonnet 4.6’s 58.1% and closing in on Opus 4.8’s 69.2%. On broader knowledge work, Anthropic says Sonnet 5 slightly outperforms Opus 4.8, which is remarkable given the price gap between the two models.
The practical takeaway is that a mid-tier model is now good enough to handle work that used to demand a flagship. That is exactly the kind of shift that makes agents viable in production. A sales team automating lead qualification, an operations team reconciling data across systems, or a support team triaging tickets can now run those workflows on a model that is both capable and affordable enough to use at scale.
Anthropic also emphasized safety improvements. Sonnet 5 shows lower rates of undesirable behavior than Sonnet 4.6, is better at refusing malicious requests, and is more resistant to prompt-injection attacks, one of the most serious risks for any agent that touches the open web or untrusted inputs. It also hallucinates less and is less sycophantic, two failure modes that quietly erode trust in automated systems. For teams deploying agents that act on real data, those reliability gains matter as much as the benchmark scores.
For sales teams tired of cold leads, slow customer responses, and manual processes, Dapta is the ultimate tool.
Dapta is the leading platform for creating AI sales agents specifically designed to increase inbound lead conversion. Respond to your leads in less than a minute with voice AI and WhatsApp that converts.
If you want your team to sell more while AI handles the complex stuff, you have to try it.
The reception from developers has been telling. Daniel Shepard, an engineer at Zapier, summed up the mood by saying that for day-to-day automation, Sonnet 5 is a no-brainer. That phrase captures the strategic shift underway. The competition among AI labs used to be about who had the most capable model. Increasingly, it is about who can deliver enough capability at a price that makes running agents at volume actually pencil out. When a model can plan, browse, and execute multi-step tasks for a couple of dollars per million tokens, the barrier to automation stops being technical and starts being about imagination.
For businesses in LATAM and other cost-sensitive markets, this is especially significant. The expensive part of AI has never been a single query, it is running thousands or millions of them as an agent works through real workflows. Cutting the per-token cost while raising capability is what turns a promising demo into a deployable system. Sonnet 5 is a clear signal that the agent era is moving from experiments to everyday operations, and that price, not just raw intelligence, is now the battleground. The teams that win will be the ones that figure out which workflows to hand to an agent first.
