San Francisco-based AI startup Anthropic has unveiled Claude Sonnet 4.5, its most advanced model yet, positioning it as a potential replacement for enterprise apps like Slack by autonomously coding, running workflows, and sustaining long tasks without human intervention.
The model has demonstrated state-of-the-art results on coding benchmarks, scoring 77.2% on SWE-bench Verified — a leading software engineering test — outperforming OpenAI’s GPT-5 on the same benchmark.
With additional parallel compute, the model can reach 82% accuracy. On Terminal-bench, another rigorous evaluation, it achieved a 50% score, which Anthropic claims is industry-leading.
What makes Claude Sonnet 4.5 stand out is its ability to focus for over 30 consecutive hours on complex, multi-step tasks — a significant leap in AI’s long-term reasoning and coding stamina. Its enhanced Computer Use feature and Agent SDK enable enterprises to design agentic workflows, automating internal processes and potentially replicating tools like Slack.
The launch comes just seven weeks after OpenAI’s GPT-5 debut, reflecting the intense rivalry in the AI space as firms battle for enterprise adoption. Elon Musk recently dismissed Anthropic’s chances on X, saying “winning was never in the set of possible outcomes.” The company declined to comment.
Still, the timing underscores Anthropic’s push to expand its customer base beyond its two core clients, showing investors and enterprises that it remains a formidable contender in the AI race.
See What’s Next in Tech With the Fast Forward Newsletter
Tweets From @varindiamag
Nothing to see here - yet
When they Tweet, their Tweets will show up here.
