Opus 5.5 (AI model): news timeline & CVEs

Exa Launches Agent Ultra: A Subagent Swarm Deep Research API Built for Exhaustive List Building

Exa launched Agent Ultra, a hosted subagent research API it says leads four deep-research benchmarks.

Exa released Agent Ultra, the highest-effort mode of its hosted Exa Agent API, for exhaustive list building, entity enrichment, and multi-source research. It splits tasks across parallel subagents and mixes frontier and faster models, invoked by setting effort to ultra. Exa’s vendor-reported results are WANDR soft recall 81.4%, DeepSearchQA F1 93.9%, WideSearch row-level F1 58.9%, and 2,451 passing entities per Company Find-All task, ahead of Opus 5.5, GPT-6 Astra, and Perplexity Agent. Typical runs take about 30 minutes, very hard tasks up to three hours, at up to $20 per run by default.

MarkTechPostSep 26, 08:04 UTC · 1d agoAI tools & infra

Will Meta’s AI Tamagotchi bet is…working?

Meta's personal AI agent Muse is reportedly outpacing ChatGPT's early growth as OpenAI and Anthropic ship new models.

TechCrunch's Equity podcast discusses Meta's consumer AI strategy after Anthropic released Opus 5.5 and OpenAI followed 90 minutes later with GPT-6 updates. Meta's personal agent Muse is reportedly outpacing ChatGPT's early numbers and is headed for smart glasses and a small Tamagotchi-style device. The hosts also cover implications for startups, a16z's school for founders, and two AI and hardware deals.

TechCrunch · AIupdated Sep 25, 18:22 UTC · 1d agofirst Sep 25, 16:00 UTC · 1d agoAI industry 2 sources

Using LLMs to trace alchemical knowledge and decode 17th century letters

A historian says frontier models traced alchemical sources and broke an unsolved 1941 cipher.

After the releases of GPT-6 Sol and Opus 5.5, a historian reports using frontier models on open historical problems rather than transcription alone. GPT-6 Astra identified a French alchemical source for a Latin passage Isaac Newton had translated, and researchers say it deciphered a 10 July 1941 German message by connecting Bundesarchiv radio-message collections that specialists had not fully linked. The author argues the useful problems are digitized, checkable, and suited to multilingual, mathematical, or cryptographic reasoning, and calls for labs and funders to support collaborations with historians. An attempt to interpret John Dee's coded Liber Loagaeth is described as inconclusive in the excerpt.

Hacker News · securitySep 24, 19:14 UTC · 2d agoAI research HN 21↑ · 4 comments

How we made claude.ai 3x faster in two weeks

Anthropic cut core claude.ai and desktop latency about 3x in two weeks, using Claude to find and ship fixes.

Anthropic engineers say they made the core experience of claude.ai and the Claude desktop app about three times faster during a two-week August sprint. At the 75th percentile, time to a typeable page on a fresh load fell from 3.1 seconds to 0.55 seconds, starting a Claude Code session fell from 0.8 to 0.3 seconds, and loading a Claude Cowork cloud session fell from 2.6 to 0.73 seconds. They focused on four journeys covering about 95% of activity and estimate the changes save tens of thousands of user-hours of waiting each day. Using Claude Tag with an internal model roughly comparable to Opus 5.5, the team merged more than 3,000 changes without a customer-facing incident or rollback.

Hacker News · AISep 23, 19:23 UTC · 3d agoAI industry 9 sources HN 53↑ · 14 comments

Opus 5.5

Timeline4 articles

Appears with

Exa Launches Agent Ultra: A Subagent Swarm Deep Research API Built for Exhaustive List Building

Will Meta’s AI Tamagotchi bet is…working?

Using LLMs to trace alchemical knowledge and decode 17th century letters

How we made claude.ai 3x faster in two weeks