OpenAI releases 722 math manuscripts from internal frontier model
Read original source →openai.com
OpenAI published 722 manuscripts (719 after three withdrawals) organized into 372 families, covering 90 of the top 500 open problems per Proof of Atlas, with Lean proof formalizations on GitHub. The single model averaged three hours of compute per solution after being asked to attempt about 4,000 problems.
People · Terence Tao
How this story unfolded
7 days · 12 reports · 43 community posts · 55 of 67 shown
- Oct 6
- Oct 7
[AINews] Quasi-Riemann-Hypothesis: OpenAI publishes 722 math papers solving 90 of the top 500 open math problems; “the most significant moment” in >100 years of mathematicslatent.space
Complementary remarks from Gary Marcus and Terence Tao on OpenAI’s giant math dropgarymarcus.substack.com
The Mathocalypsescottaaronson.blog
OpenAI Says a Secret AI Model Cracked Hundreds of Open Math Problems in One Prompt—Mathematicians Want Receiptsdecrypt.co
- Oct 8
- Oct 9
- Oct 10
- Oct 11
- Oct 13
More stories today
Scoble: AI agents merging their own PRs at hundreds of deploys a day
Robert Scoble·2 hours ago
Meta: Muse Wonder won't be open-sourced, Muse Spark weights coming soon
Meta says Muse Wonder models are too large for many people to use and carry potential safety concerns, so they will not be open-sourced. Open weights for Muse Spark are coming soon.
AI at Meta·2 hours agoAnthropic builds Scion, in-house version control, as merges hit 30k/day
Anthropic migrated its monorepo off git to an in-house version control system called Scion in five months, after daily merges grew from ~2,500 in May 2026 to over 30,000 by September. It also shipped a code review surface called Antique, code search, and a source control IAM system in the same period.
Anthropic Engineering·2 hours ago

LangChain moves MCP support into langchain.mcp for new spec
MCP support now ships in the main langchain.mcp package, built on FastMCP for the 2026-07-28 spec, replacing the separate langchain-mcp-adapters install. Elicitation is surfaced as a LangGraph interrupt and tool lists are now cacheable, matching the spec's stateless core.
LangChain Blog·2 hours ago

LangChain: open models GLM-5, MiniMax M2.7 match closed frontier models on agent tasks
LangChain's Deep Agents evals found GLM-5 (z.ai) and MiniMax M2.7 score similarly to closed frontier models on file operations, tool use, and instruction following. Outputting 10M tokens/day costs ~$250/day on Opus 4.6 versus ~$12/day on MiniMax M2.7 — about $87k annually.
LangChain Blog·2 hours ago

LangChain uses Deep Agents CLI to make docs code samples self-testing
LangChain automated migrating inline docs code samples into standalone, CI-tested files using its Deep Agents CLI, so broken examples fail the build. The manual path otherwise requires extracting snippets, adding setup/teardown, snippet markup, and CI wiring.
LangChain Blog·2 hours ago

Nvidia's Huang plans $2B in Europe AI deals within a week
Jensen Huang said at a Paris event that Nvidia will announce $2 billion in European investments over the next week, on a tour that also takes him to London and Berlin.
Bloomberg Technology·2 hours ago

Kai-Fu Lee: AI's opportunity is revenue, not cost savings
Kai-Fu Lee·2 hours ago