Liquid AI d1-ultra Matches GPT-6.1 with Perfect ClaimBench Score

Liquid AI's d1-ultra decision model achieved 100% accuracy on the ClaimBench financial verification benchmark, matching the performance of general-purpose models including GPT-6.1 Sol and the Claude 5.5 family. The model correctly identified all 1,960 claims on the test, which evaluates whether 10-K corporate excerpts support, contradict, or ignore specific statements. d1-ultra costs $0.21 per 1,000 claims and responds in under a second, making it roughly 12x cheaper and 5x faster than Haiku 5.5.