OpenAI's newest model is solving longstanding mathematical conjectures while a namesake token struggles with near-zero liquidity
OpenAI’s GPT-5.6 Sol Ultra is having quite the first week. The model, which launched on July 9 as part of the broader GPT-5.6 family, has already been credited with generating a complete proof for the Cycle Double Cover Conjecture, a famously unsolved problem in graph theory. It completed the task in under one hour using 64 parallel subagents.
Naturally, the crypto world responded the only way it knows how: someone launched a meme token.
What GPT-5.6 Sol Ultra actually did #
The GPT-5.6 model family includes three tiers: Terra, Luna, and the flagship Sol. The Sol Ultra mode is the heavy hitter, distinguished by its parallel multi-agent reasoning capability.
On July 10, just one day after general availability, Sol Ultra tackled the Cycle Double Cover Conjecture. For the non-mathematicians in the room: this is a problem in graph theory that has stumped researchers for decades. The conjecture asks whether every graph without a specific type of edge (called a “bridge”) can have its edges covered by a collection of cycles where each edge appears exactly twice.
The prompt and resulting proof were made publicly accessible, which is notable. OpenAI is clearly signaling confidence in the output’s validity by inviting peer scrutiny rather than keeping it behind closed doors.
Sol Ultra scored 91.9% on Terminal-Bench 2.1, a benchmark designed to measure advanced reasoning. Pricing for the Sol tier sits at $5 per million input tokens and $30 per million output tokens, with Terra and Luna offering lower price points for less demanding workloads.
The Erdős connection, and the confusion #
Here’s where things get murky. Claims have circulated that Sol Ultra solved “Erdős problem #793” using an improved construction. The Erdős problems are a legendary collection of open questions posed by Hungarian mathematician Paul Erdős, many carrying cash bounties for solutions.
What’s verified is that earlier 2026 AI models from OpenAI worked on the Erdős planar unit distance problem. There is no confirmed, independently verified connection between GPT-5.6 Sol Ultra and a specific Erdős problem numbered 793.
The Cycle Double Cover Conjecture proof is the documented achievement. Everything else is, at this point, unverified.
Enter the meme token #
A Solana-based meme token called $5.6SolUltr appeared almost immediately after the GPT-5.6 launch. The token has no disclosed utility beyond its namesake association with OpenAI’s model. It does not appear to be affiliated with OpenAI in any capacity. The token shows near-zero trading volume and negligible liquidity since inception.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our