ulam.ai 's GPT-6 Astra xhigh scored first on the ErdosBench benchmark of 226 open math problems inspired by Erdos problems.
Public source
Publisher name
Public post
GPT-6 Astra xhigh scored 1st on our ErdosBench, a benchmark of 226 open math problems inspired by Erdos problems. Interestingly, it's not that big of a leap compared to…
Company
ulam.ai
We want to make machines think
- Industry
- Information Services
- Location
- Warsaw, PL
- Company size
- 2–10 employees
About ulam.ai
ulam.ai is a company specializing in artificial intelligence. We do our own research, especially on automating creative thinking. Apart from that we offer consulting services in the form of workshops and brainstorming session to any businesses around the world.
See moreLatest activity
Latest activity from ulam.ai
3 signals
Products & Services
ulam.ai released Muse Spark 1.3, a model for mathematics that trails slightly after GLM-5.3 Flash and GPT-5.6 Sol.
Products & Services
ulam.ai released 1,000+ downloads of its AIME++ sample data, which includes high-school level math, graduate math, and research level problems encoded in 0-999 integer answer format.
Discover more
Similar signals
Similar public activity from other companies.
Research & Knowledge
Epoch AI
Epoch AI set the top score on the FrontierMath Erdős math benchmark, consisting of 68 unsolved and especially interesting Erdős problems curated by Thomas Bloom, scoring 3%.
Research & Knowledge
OpenAI
OpenAI shared a solution to the Navier-Stokes Millenium Prize Problem, produced by a group of agents using an OpenAI next-generation model significantly more capable than GPT-6 Astra.
Products & Services
iSolutionsAI
iSolutionsAI's GPT-6 Astra saturates FrontierMath Tier 4 with a 98% score.
Research & Knowledge
Vals AI
Vals AI reports that Astra scored 99% on ProofBench, solving all but one task (measures solving grad-level math proofs in Lean).
Research & Knowledge
Boston Consulting Group (BCG)