Palo Alto-based AI startup Pathway has raised fresh funding at a valuation of $500 million, taking its total seed funding to $30 million. The company plans to use the money to expand its computing capacity, including new Nvidia GB300 systems, and develop its "Post-Transformer" architecture. Investors in the round include Id4 Ventures, TQ Ventures, Red Bridge Ventures, Kadmos Capital and WS Investment Co., with Jonathan Frankle, chief AI scientist at Databricks, participating as an angel investor. Alongside the funding, Pathway released test results for BDH-CQ, a reasoning model with 150 million parameters, which scored 29.5% on ARC-AGI-1, a benchmark testing an AI system's ability to figure out a rule from a few examples. Pathway tested its model across 400 publicly available tasks under the benchmark's "pass@2" system, allowing two attempts per problem. Pathway said each task cost about $0.0007 in computing resources, making BDH-CQ about 11 times cheaper per task than GPT-5.6 Luna (Low), which achieved a higher score of 34.2% on the same benchmark. That comparison includes an 80% reduction in GPT-5.6 Luna's public API prices announced on July 30; using July prices before the reduction, Pathway said its model would have been about 57 times cheaper. Pathway argues its architecture handles reasoning differently, reducing the amount of computing power