What happened
OpenAI and Anthropic are being compared across frontier artificial intelligence benchmarks, especially software-engineering and agentic-workflow tests. The comparison does not produce a single winner: Anthropic’s Fable 5 remains stronger on SWE-Bench Pro, while OpenAI’s GPT-5.6 Sol claims advantages on some other agentic benchmarks and on efficiency.
Background and earlier position
Anthropic gained an early rollout advantage through wider user access to its Claude-led systems, later widened by Fable 5 deployment. Access to specific model versions was paused for 19 days after a dispute with the U.S. government over export controls before restoration.
OpenAI and Anthropic were compared using benchmark results that reward different skills. Some tests measure precise bug-fixing inside a codebase, while others reward broader computer-operation ability, including setup, execution, recovery from errors, and multi-step workflow completion.
Related current affairs
- Unprecedented: OpenAI system acts on its own and hacks computer
- Jitin Prasada holds bilateral meetings in Morocco on trade, investment and digital transformation
- What changes when AI moves from reading viral genomes to designing them?
- What does Anthropic stand to gain and risk with its IPO?
- Google’s AI advantage faces its biggest test yet
- AI speech recognition model trained on 65 Indian languages and dialects released