Trajectory launches platform for post-training trillion-parameter models, raises $60M led by Sequoia at $300M valuation

Trajectory has launched a platform it describes as a tool for post-training trillion-parameter models, offering a suite to train, evaluate, and deploy them via agent prompts. Two research previews accompany the release: Continual Grading, which scales user preferences beyond manual grading, and GS1, an algorithm letting models learn from their own production behavior. The company says it raised $60M led by Sequoia at a $300M valuation, with backing from Conviction, NVIDIA, Bessemer, Bain Capital, Pear VC and others. Access is via a waitlist.

✨ AI summary of the linked source — not editorial content.

Explore the topic

• Launch
• Raises