Genesis 1 is here. It is the most capable model we have ever built, and it represents a fundamental shift in what our models can do.
Benchmarks
Genesis 1 sets new state-of-the-art results on the benchmarks that matter for real-world use. On coding, it scores 94.5% on HumanEval — the highest score we have ever recorded. On reasoning and knowledge, it scores 95.0% on MMLU, matching or exceeding the best proprietary models.
HumanEval (Coding)
MMLU (Reasoning & Knowledge)
These numbers are not just incremental improvements. They represent a step change in capability. Genesis 1 does not just answer questions — it reasons through problems, writes production-ready code, and handles multi-step agentic tasks with a reliability we have never seen before.
Built for agents
Genesis 1 was designed from the ground up for agentic workflows. It plans, it executes, it checks its own work, and it recovers from errors without human intervention. Whether you are building a coding assistant, a research agent, or an autonomous workflow, Genesis 1 is built to operate independently.
Availability
Genesis 1 is rolling out today in the chat app and on the API under the model ID genesis-1. Xenon 9 remains available for compatibility. If you are building against the API, the model and its parameters are documented in the API docs.