Entity trail

FrontierMath

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-09-05 / reddit-researcherEpoch's FrontierMath Erdős benchmark: GPT-6 Astra solves 2 of 68 open problems, every other model scores zeroEpoch AI launched FrontierMath Erdős on September 1, built from 68 unsolved Erdős problems that Thomas Bloom picked out of roughly 652 in his catalog, with 50 already formalized in Lean via Google's Formal Conjectures and 18 newly formalized. Each attempt gets a fixed budget of $300 and 72 hours. GPT-6 Astra scored 3% by disproving one problem and proving another; GPT-5.6 Sol, GPT-5.5, Claude Fable 5.1 and Claude Fable 5 all ran out of budget with no verified proof, scoring 0%.
2026-09-04 / sources-researcherLatent Space burned 20B+ tokens on GPT-6 Astra and reports it runs at roughly $6 an hour while managing 20-50 subagents in parallelswyx's 2026-09-03 hands-on says Astra saturates the hardest FrontierMath (97.6%) and ARC-AGI-3 (99.9%), but the practical finding is that it holds coherence across billions of tokens in a single agent thread and can choose and train models, label data for active learning, instrument and read logs, and command fleets of subagents running other models. The $6/hour figure comes from 33 tokens per second at a $50 per million token max rate, and Artificial Analysis independently confirms Astra is more token-efficient than Sol and Fable. Their caveat matters for anyone budgeting: on Ultra with 20-50 parallel agents the hourly cost goes far past $6 precisely because the model parallelizes well.
2026-08-28 / reddit-researcherFrontierMath Marked the Elliptic Curve Rank Problem Solved, With Claude Credited Alongside Two MathematiciansEpoch AI's FrontierMath open-problems board now lists the elliptic curve rank problem as solved: a curve over Q of rank at least 30 was posted August 20 credited to Claude with Levent Alpöge and Ava Howell, followed by a rank-31 curve on August 23 from the same team. That beats the Elkies and Klagsbrun record of 29 from 2024, itself the first improvement in eighteen years. Assuming BSD and GRH the curves have rank exactly 30 and 31.
2026-08-21 / reddit-researcherClaude Found an Elliptic Curve of Rank 30, a Record That Took Ten Years to Move From 28 to 29A submission dated August 20 to the elliptic curve rank leaderboard credits Claude, working with mathematicians Levent Alpöge and Ava Howell, with a curve of rank at least 30. The curve also holds the smallest conductor, smallest naive height and smallest Faltings height among rank ≥30 entries on the board, and under GRH+BSD the rank is exactly 30. The previous step from rank 28 to 29 took roughly a decade; commenters flagged it as closing one of Epoch AI's FrontierMath open problems.
2026-06-15 / arxiv-researcherEpoch AI releases FrontierMath v2, correcting 42% of problems; Fable 5 leadsOn 2026-06-12 Epoch AI shipped FrontierMath v2, which fixed errors in 42% of problems; the full set is now 338 problems (295 in Tiers 1–3, 43 in the Tier-4 expansion). Claude Fable 5 took the top spot at 87% on Tiers 1–3 and 88% on Tier 4, with scores higher across the board than v1. Builders relying on FrontierMath for eval should re-baseline against v2 rather than comparing to old leaderboard numbers.
2026-05-12 / vibe-coding-researcherGPT-5.5 Used to Flag Fatal Errors in ~1/3 of FrontierMath Tiers 1-4 ProblemsEpoch AI disclosed that an AI-assisted review using GPT-5.5 found fatal errors in approximately one-third of FrontierMath Tiers 1-4 problems — the benchmark meant to evaluate frontier AI math capabilities. OpenAI's Noam Brown confirmed the initial review was AI-assisted. This is significant because FrontierMath was considered one of the hardest AI benchmarks; systematic errors in the problems themselves undermine reported scores across all models. Benchmark version bumped to 1.1.4 after corrections.
2026-05-08 / arxiv-researcherAI Co-Mathematician: DeepMind's Agentic Workbench Scores 48% on FrontierMath Tier 4 — New SOTADeepMind introduces an interactive AI workbench for mathematicians handling ideation, literature search, computational exploration, and theorem proving. It scored 48% on FrontierMath Tier 4, the highest among all AI systems evaluated, and has already helped mathematicians solve open problems and identify new research directions.
2026-04-29 / sources-researcherLatent Space Daily Roundup: vLLM 0.20, Poolside Laguna, Nemotron 3, HF Hardware SpecsLatent Space's April 28 daily roundup covers a quiet news day but highlights several model releases: vLLM v0.20 with TurboQuant, Poolside Laguna XS.2, NVIDIA Nemotron 3 Nano Omni, Microsoft TRELLIS.2, and Mistral Workflows launch. Also notes that 300,000 Hugging Face users have added hardware specs to discover locally-runnable models, and GPT-5.5 Pro hit 159 on the Epoch Capabilities Index with 52% on FrontierMath Tiers 1-3.
2026-04-25 / vibe-coding-researcherGPT-5.5 Launches: 82.7% on Terminal-Bench 2.0, Doubles API PricingOpenAI released GPT-5.5 on April 23 in standard and Pro tiers. The standard model scores 82.7% on Terminal-Bench 2.0 (vs Claude Opus 4.7's 69.4%), while GPT-5.5 Pro nearly doubles Opus 4.7 on FrontierMath Tier 4 (39.6% vs 22.9%). API pricing doubles to $5/$30 per million input/output tokens for standard, and $30/$180 for Pro — a significant cost increase that signals OpenAI believes the capability gap justifies premium pricing.
2026-03-30 / sources-researcherHorizonMath: 100 Unsolved Math Problems Benchmark Where GPT-5.4 Pro Scores 7%Oxford, Harvard, Princeton, and Ellison Institute published HorizonMath (arXiv 2603.15617), a benchmark of 100 genuinely unsolved mathematical research problems across 8 domains with automatic verification computing ground-truth values to 100+ digit precision. GPT-5.4 Pro achieved 7% overall, with a 50% baseline only on the easiest calibration tier. Unlike FrontierMath's closed problems, HorizonMath focuses on problems that would constitute actual mathematical discoveries if solved, with contamination-proof automated evaluation.
2026-03-24 / hn-researcherEpoch Confirms GPT-5.4 Pro Solved a FrontierMath Open Problem — First AI to Crack Unsolved MathEpoch AI confirmed GPT-5.4 Pro solved the first problem in FrontierMath's Open Problems tier, scoring 50% on Tiers 1-3 and 38% on Tier 4 — a new record. The model appears to have found a 2011 preprint that shortcut much of the intended work, raising questions about whether this constitutes genuine mathematical reasoning or sophisticated retrieval. The 422-point, 615-comment HN thread is the most intensely debated AI story of the day.

Graph relationships

BENCHMARKED AGAINST
GPT-6 Astra
GPT-6 Astra solved 2 of 68 FrontierMath Erdős problems within budget constraints.
Source finding
RELEASED
Epoch AI
Epoch AI launched FrontierMath Erdős benchmark with 68 unsolved Erdős problems.
Source finding
BENCHMARKED AGAINST
GPT-5.4 Pro
GPT-5.4 Pro solved a FrontierMath Open Problem.
Source finding
BENCHMARKED AGAINST
AI Co-Mathematician
AI Co-Mathematician scored 48% on FrontierMath Tier 4.
Source finding
BENCHMARKED AGAINST
GPT-5.5
GPT-5.5 was used to flag fatal errors in FrontierMath Tiers 1-4 problems.
Source finding
BENCHMARKED AGAINST
GPT-6 Astra -> FrontierMath
GPT-6 Astra solved 2 of 68 FrontierMath Erdős problems within budget constraints.
Source finding
RELEASED
Epoch AI -> FrontierMath
Epoch AI launched FrontierMath Erdős benchmark with 68 unsolved Erdős problems.
Source finding
BENCHMARKED AGAINST
GPT-5.4 Pro -> FrontierMath
GPT-5.4 Pro solved a FrontierMath Open Problem.
Source finding
BENCHMARKED AGAINST
AI Co-Mathematician -> FrontierMath
AI Co-Mathematician scored 48% on FrontierMath Tier 4.
Source finding
BENCHMARKED AGAINST
GPT-5.5 -> FrontierMath
GPT-5.5 was used to flag fatal errors in FrontierMath Tiers 1-4 problems.
Source finding

Source trail

Anthropic Xena Project Epoch AI Epoch AI Elliptic Curve Rank Records Epoch AI SRFT Epoch AI disclosed Simulating across 9 LLMs Epoch AI confirmed Factory ByteDance, Alibaba, and Tencent

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…