Grok 4.6 멀티모달 활용, 기업 에이전트 맥락 그래프, 다중 벡터 RAG와 잠재 추론

Elon Musk

@elonmusk

Grok 4.6

Grok 4.6 multimodal is a step change from Grok 4.5. It’s one of the under-discussed improvements, and I’ve been very impressed by it. My daily work includes reviewing lots of videos and understanding the context; Grok 4.6 improves the productivity of such workloads by at least 10x if not 100x. Such workflows may not be captured by common VLM benchmarks, but in my use cases it outperforms Gemini 3.5 and Gemma 4, which is considered the SOTA of VLMs in my opinion. Hats off to the multimodal teams—you did a great job.

Ali Ghodsi

@alighodsi

I got this question so many times today. "How can you grow 80% at $7B?" The true answer is that we're finally seeing a breakthrough with AI agents starting to work in the enterprise. The AIs have been super smart for a while, but have lacked basic context that's in people's heads, or in some SaaS system-or-record. A lot of organizations are deploying FDEs to capture this context, or Ontology, and feed it to the AI. This is labor intensive and expensive. We just automated that with Genie Ontology. Once you have that enterprise context graph, an AI agent like Genie becomes magical. I find myself no longer waiting for answers from my CRO, CFO, CMO, CHRO etc, I just keep queuing up questions on the phone while sitting in meetings. It'd frankly addictive. Our customers are starting to do the same, over 70% of all queries on the platform are now generated by Genie agents. This fuels more questions to the platform, which drives consumption, which drives revenue. That's the simple answer.

This morning, Databricks Co-Founder and CEO @alighodsi joined CNBC's @jonfortt to discuss our recent funding and how Lakebase, our serverless Postgres database built for AI agents, along with Genie and Unity AI Gateway, give organizations the context, control and choice they need to scale AI. “During big transitions like this, it’s good to be able to just focus on your business. If there is so much demand, just build for the future,” said Ghodsi. Watch the full conversation: https:// cnbc.com/2026/08/13/dat abricks-funding-round-190-billion-valuation.html …

YanXbt

@IBuzovskyi

HERMES AGENT NOW HAS BOT MODE. YOUR AGENT PROFILES BECOME A ROSTER OF NAMED BOTS. EACH WITH ITS OWN CHAT, AVATAR, AND PERSONALITY. THEY MESSAGE EACH OTHER. sessions mode: you open chats. you manage profiles. you switch between agents manually. bot mode: a roster of bots in the sidebar. click one. you're in its chat. each bot has a name, an avatar, a description, its own memory, skills, and SOUL.md. they can talk to each other. WHAT YOU GET: BOTS PANE: left sidebar shows every agent profile as a bot. avatar, latest message preview, timestamp. click a bot to open its chat. no profile switching. CREATE A BOT: name, title, description. done. advanced: clone from existing profile, pin a model, write SOUL.md, toggle skills. EDIT ANY TIME: right-click a bot. change avatar, title, description. edit its model, skills, toolsets, and SOUL.md live. DUPLICATE: right-click. full clone. config, skills, SOUL.md, memory. everything. AVATARS: geometric faces (7 shapes, 10 colors, blinking eyes that scan while the bot works). upload your own image. or AI-generated portrait (when image backend is set). pixel pet companion bounces beside the avatar while the bot is busy. ROUTINES: recurring tasks per bot. backed by Hermes cron. "summarize my inbox every morning" lives next to the bot that does it. runs land in the bot's own chat history. BOT-TO-BOT MESSAGING: every bot has a persistent Agent Inbox. bots message each other with attribution: "[Message from agent 'researcher']" their SOUL.md teaches them the protocol. including how to reply. @ MENTIONS: type @ researcher in any chat. the active bot hands the message off. waits for the reply. reports back. "@ researcher have a look at this" → researcher reads the message → researcher responds → active bot delivers the response to you HOW IT WORKS UNDER THE HOOD: a bot IS a Hermes profile. isolated config, memory, skills, credentials. lives at ~/.hermes/profiles/<name>/ bot mode is a UI over that primitive. no core patches. no background daemons. no extra storage. everything is standard Hermes. bot-to-bot messages = real CLI handoffs: hermes -p <bot> chat -c "Agent Inbox" -q "..." routines = plain Hermes cron jobs namespaced [bot:<name>] <routine>. show up in hermes cron list too. INSTALL: git clone https:// github.com/NousResearch/H ermes-Bot-Mode … ~/.hermes/desktop-plugins/hermes-bots reload plugins: Ctrl+K → "Reload desktop plugins" or restart the app. a Bots tab appears next to Sessions. a Routines tile docks beside the conversation. PUBLIC BETA: this is a one-day public beta test. install it. try it. report bugs. feedback goes to the GitHub repo. then it ships in the main Desktop app for everyone. http:// github.com/NousResearch/H ermes-Bot-Mode … P.S. If you noticed there is a pet of Rebecca from cyberpunk edgerunners will soon finish and drop a set of characters. Any fans of Cyberpunk in Hermes community?)

Claude Code CLI 2.1.232 changelog: New features: • Added /config rows for "Dialog expiry" and "Messages from your other sessions" (cross-session inbound accept/hold/refuse) • Added secret redaction for GitLab token families (glrt-, gloas-, glptt-, glagent-, glimt-, glsoat-, glcbt-, glft-, glffct-) and full redaction of routable glpat-/gldt- tokens; the glab CLI config store gets the same sandbox and credential-path protection as gh • Added GitLab support to plugin marketplaces: bare http:// gitlab.com repo URLs (including nested subgroups) now clone like http:// github.com URLs, and clone auth-failure hints name your actual git host Fixes: • Fixed a PowerShell permission bypass where variable-writing parameters could silently overwrite ＄PSDefaultParameterValues and redirect later commands' file access • Fixed a Windows permission bypass where Git Bash followed Cygwin-style symlinks that path validation saw as regular files; writes through them now require permission approval • Fixed nested git repositories inheriting trust from a parent directory; each repository now requires its own trust confirmation • Fixed MCP connections hanging for the full 30-second connect timeout when a server fails to answer or sends a malformed reply to the protocol-version probe • Fixed Remote Control sessions hosted by a bridge inside a cloud session inheriting that session's transcript or credentials • Fixed Remote Control sessions started from Claude Desktop or an IDE appearing as a new http:// claude.ai session each time the local session was resumed; they now reattach to the existing one • Fixed Remote Control sessions appearing unreachable to newly attached clients while idle • Fixed Remote Control bridge sessions not restoring conversation history when the session worker restarts • Fixed Cloud gateway /login exiting silently or leaving an unresponsive terminal after "Press Enter to continue" when managed settings failed to load; the reason is now shown • Fixed voice mode on native builds getting stuck on "listening…" when the voice service rejected the connection; the rejection is now shown immediately • Fixed mTLS client certificate rotation requiring a restart; Claude Code now reloads the rotated cert and key automatically on connection errors • Fixed malformed AWS or Vertex region values being used to build request URLs; they now fall back to the default region • Fixed stream idle timeout errors failing the request instead of recovering on Bedrock, Vertex, and gateway deployments • Fixed content-sized overlays containing truncated text rendering one column too wide, and start-truncated text collapsing to an ellipsis • Fixed a stray garbled character where a long shell-command or agent-description preview was cut off mid-emoji • Fixed a startup race that could silently unregister a plugin marketplace due to concurrent writes to known_marketplaces.json • Fixed /update and /tui refusing to restart while work that survives the relaunch was running • Fixed usage-limit guidance suggesting unavailable slash commands in SDK and remote sessions • Fixed the consent message for interactive --advisor fable launches, which told you to run /model fable in an interactive session that had just exited Improvements: • Improved fullscreen streaming: long sessions stay responsive because the whole conversation is no longer re-normalized on every update • Improved the managed settings approval dialog: shows endpoint URLs, uses clearer wording for telemetry-only changes, skips routine OpenTelemetry options, and requires approval for server-managed sandbox binary overrides (sandbox.bwrapPath, sandbox.socatPath, sandbox.ripgrep) Removal: • Removed the startup tip suggesting you create custom subagents, and the matching nudge in the /powerup tour Security/safety changes: • Bash input redirections (< file) are now permission-checked like their argument spellings on all platforms • Hardened the Linux filesystem sandbox against a protected-path bypass • Changed sandbox.ripgrep to be honored only from user, managed, and --settings settings; project settings can no longer override the sandbox's ripgrep binary Other changes: • Subagent forking is now on by default: a subagent_type: "fork" subagent inherits the full conversation and prompt cache, and non-teammate agent spawns in interactive sessions now run in the background by default • Type @ in the prompt to mention another Claude session by name; Claude then uses SendMessage to reach that session directly • SendMessage now delivers to a bare name that exactly matches one live session, instead of asking to confirm with a ref first • Interactive sessions on one machine now keep unique names: starting or renaming a session to a name another live session already uses gives it a name-word-word variant and tells you • Settings: additionalMarketplaces and allowedMarketplaces are now accepted as friendlier aliases for extraKnownMarketplaces and strictKnownMarketplaces • Enterprise policy: a url-typed blockedMarketplaces entry for a bare repo URL keeps blocking that URL when the CLI classifies it as a git clone • Gateway: the desktop: overlay now accepts every released Desktop setting (was 11 hand-listed keys), validated at boot against Desktop's own schema; unknown or invalid keys fail boot • Gateway: empty managed.policies[].match.groups/admin.admin_groups entries and malformed email_domain values (empty, or containing @, whitespace, or commas) now fail at boot instead of silently matching no one or granting admin access • Fable 5 is offered as an advisor in /advisor again for organizations with Fable access, with usage-credits consent set up through /model fable • Remote Control: resuming a conversation whose session was deleted from http:// claude.ai or the app now starts a replacement instead of failing with a message about your login (regressed in v2.1.227) • /feedback and /bug now open immediately when invoked while Claude is responding, instead of waiting for the turn to finish • /plugin install plugin@marketplace now refreshes the marketplace first, so newly published plugins install without a manual marketplace update • /code-review at high, xhigh, and max effort now runs in a background agent like the other levels • Pasted and clipboard images are read without blocking the event loop • Remote Control now keeps reconnecting for about 30 minutes after a network blip and no longer drops after a few blips spread across an hour • Remote Control: resuming a conversation no longer silently takes Remote Control away from another Claude Code on the same machine that still has it; run /remote-control there to move it • Updated agent panel: completed subagents hide immediately with a /tasks footer hint, and the "↓ N more" overflow indicator moved left for visibility • Remote Control: the terminal now says whether a session was taken over by another device, ended from another app, or deleted, and stops suggesting a reconnect that would undo it • Shortened the message shown when resuming a completed background agent • Cowork sessions no longer inline external @-imports from user-scope memory files • Hardened the auto-generated cross-session messaging socket directory on shared /tmp: a pre-planted symlink or another user's directory is now refused instead of used Source: https:// github.com/anthropics/cla ude-code/blob/main/CHANGELOG.md#21232 …

When one vector is not enough for RAG Single-vector retrieval works well for many workflows. But when documents get longer, denser, or visually complex, compressing everything into one embedding can lose important context. A long technical report, a PDF full of tables, a slide deck, or a scanned document often contains meaning across tokens, layout, charts, and visual structure. That is where multi-vector retrieval changes the game. In our latest blog, we break down how Langflow 1.11 introduces first-class multi-vector retrieval with the new lfx-nextplaid extension bundle. With NextPlaid, Langflow flows can now use ColBERT-style late interaction and ColPali-style visual document retrieval without custom glue code. What this enables: - Token-level retrieval instead of relying on a single pooled document vector - Better search quality for long and information-dense documents - Visual retrieval for PDFs, tables, charts, slide decks, and scanned pages - PLAID indexing for faster, more memory-efficient multi-vector retrieval - New NextPlaid and vLLM Multivector Embeddings components inside Langflow The benchmark results are significant. On long technical reports, text multi-vector retrieval via PLAID reached 94.7 Recall@10, compared to 73.7 with single-vector retrieval. For image retrieval, multi-vector retrieval reached 89.7, compared to 22.6 with a single-vector baseline. If you build RAG workflows over complex documents, technical reports, enterprise knowledge, or visual PDFs, this deep dive is worth reading. Read the full blog: https:// langflow.org/blog/blog-next plaid?utm_source=social&utm_medium=organic&utm_campaign=nextplaid_blog …

ChatGPT

@ChatGPT

You probably hate tab switching. Understandable. You can now open any @googledrive Doc, Sheet, or Slide right inside ChatGPT and work side by side without switching tabs. Rolling out on web to Plus, Pro, Business, and Enterprise users in ChatGPT and ChatGPT Work.

“BDH-CQ: In-Context Learning with Recurrent Latent Reasoning” Instead of generating CoT, this paper lets examples update a recurrent memory, then repeatedly reasons inside a continuous latent state without verbalizing intermediate steps. A 150M model reaches 29.5% on ARC-AGI-1 for just $0.0007 per task, setting a new cost-efficiency frontier for the benchmark. https:// alphaxiv.org/abs/2608.09888 v1 …

Robots that predict the next world state before they move can adapt to tasks they've never seen. That's World Action Models and we'll be demoing them live 8/26, plus new eval tables in W&B Models and @CoreWeave ARIA (AI Research Iteration Agent) message queueing. https:// utm.io/ur8ga