Pins matching tag:"Computer Use"

Anthropic releases an upgraded Claude 3.5 Sonnet that lifts SWE-bench Verified from 33.4% to 49.0% and is the first frontier model to offer computer use, moving a cursor and typing like a person, in public beta.

Anthropic releases Claude Sonnet 4.5, which it calls the best coding model in the world at 77.2% on SWE-bench Verified, leads OSWorld computer use at 61.4% and keeps working on one task for more than 30 hours.

Anthropic releases Claude Haiku 4.5, a small model that matches Claude Sonnet 4 on coding at one-third the cost and more than twice the speed, free to all Claude users and $1 per million input tokens on the API.

Anthropic releases Claude Sonnet 4.6, its most capable Sonnet, which coding testers preferred to Opus 4.5 and which becomes the free default in claude.ai with a 1M-token context window at $3 per million input tokens.

Cursor's cloud agents gain their own isolated virtual machines where they run the software they build, test changes and hand back merge-ready PRs with video, screenshot and log artifacts.

Alibaba releases Qwen3.7-Plus, a proprietary multimodal agent model that reads images and video, operates screens and scores 79.0 on ScreenSpot Pro, priced at about $0.40 per 1M input tokens.

Anthropic releases Claude Sonnet 5, its most agentic Sonnet, which comes close to Opus 4.8 on coding and computer use at $2 per million input tokens and becomes the default model for Free and Pro users.

OpenAI releases GPT-6 Astra, which it calls the world's most intelligent and aligned model, first to a limited set of organizations and then to paid ChatGPT plans, the API and AWS; it is the first OpenAI model rated Critical for cybersecurity.

OpenAI releases GPT-6.1 Sol for coding, computer use and professional work, with API input and output prices one-fifth of GPT-6 Astra’s and lower cached-input pricing.