Two major artificial intelligence (AI) companies, OpenAI and Anthropic introduced new flagship models on February 5.
Anthropic introduced Claude Opus 4.6. The company said the model handles large volumes of text more effectively and supports agent-style workflows.
It highlighted a 1-million-token context window and a 76% result on the MRCR v2 benchmark. This test measures how well a system retrieves and interprets complex information.
Did you know?
Subscribe – We publish new crypto explainer videos every week!
What is an Automated Market Maker in Crypto? (Animated)
Anthropic also stated that Claude Opus 4.6 improves on earlier versions in finance and legal tasks. It added a feature called “agent teams”, which lets several AI agents work at the same time on coding or documentation tasks.
OpenAI released GPT-5.3 Codex later. The company presented it as a model built for agentic coding and research tasks.
OpenAI said Codex reached 77.3% on Terminal-Bench 2.0, which measures performance on automated coding tasks. The company also reported that Codex completed tasks faster and used fewer tokens.
OpenAI added that early versions of Codex were used for internal work. The company used the model to help debug training processes and manage deployment.
The initial results showed that the two models focus on different strengths. Claude Opus 4.6 performed better on legal and financial reasoning. GPT-5.3 Codex delivered stronger results on coding tests and overall efficiency.
Meanwhile, Elon Musk recently decided to bring xAI under the control of his space company, SpaceX. What did he say? Read the full story.
