The Gradient / Index
The reading room.
Search the stories, research, and sources from every issue.
93 editions · 461 stories
93 issues found
Introducing ChatGPT for Financial Services
OpenAI has unveiled a finance‑focused ChatGPT, merging its proprietary market data with the new GPT‑6 Astra engine to streamline research, modeling, and client deliverables.
GPT-6 Astra: The next generation in intelligence for work
OpenAI’s latest model, GPT‑6 Astra, is engineered to boost productivity with sharper reasoning, computer‑tool use, and refined writing and design judgment.
How GPT-5.6 Sol helps run quantum computing experiments
A breakthrough at MIT shows how the latest GPT-5.6 Sol model, paired with Codex, can run quantum computing experiments end-to-end without human intervention.
Hikers rescued after using Google Gemini for planning
A weekend trek turned into a rescue when hikers followed Google Gemini’s planning tips, which underestimated essential supplies. The incident underscores the dangers of relying solely on AI for outdoor safety.
From Matching Models to Recruiting Agents: A Systematized Narrative Review of AI Recruitment Systems, Evaluation, and Governance
A new systematized narrative review maps the rapid shift in AI‑driven hiring, from basic profile pairing to multi‑stage, agent‑based workflows.
Hikers rescued after using Google Gemini for planning
A weekend trek turned perilous when an AI planning tool gave hikers dangerously low provisioning advice, prompting a swift rescue.
Speculative Macro Commit for Faster Tool-Using Agents
A new runtime technique, Speculative Macro Commit (SMC), promises to slash the latency that plagues tool‑using language model agents.
Legora reviewed 41 documents in minutes with GPT-6 Astra
Legora harnessed GPT-6 Astra to slash document‑review cycles, catching every planted mistake in a high‑stakes financial workflow.
Architecting Conversational Data Systems for Stateless LLM APIs: The Hydration Proxy Pattern
Researchers unveil a new architecture that separates session persistence from large‑language‑model reasoning, easing the burden on client apps.