Ecosystem
Grok news, curated
Launches, model releases, and builder-tool moves - summarized in one line each, always linked to the original source.
- SpaceXSI News
Grok Voice 3.0 tops the speech-to-speech index, same API price
SpaceXSI released Grok Voice 3.0, a speech-to-speech model that ranks first on Artificial Analysis's Speech-to-Speech Index at 86.9, handles interruptions well and answers in under a second on average. It is available in the Voice Agent API at the same $0.08 a minute as the previous version, and agents already pointed at grok-voice-latest were switched over automatically.
- SpaceXSI News
The X Subscription bundles Grok, Grok Bot, Cursor and X Premium
SpaceXSI put Grok, Grok Bot, Cursor and X Premium on one subscription, priced from $8 to $200 a month, with a single shared pool of Grok model usage across Grok and Cursor. Existing SuperGrok, Cursor and X Premium subscribers can keep their plan or switch and get a prorated refund.
- xAI
Team Bots: AI coworkers that learn from your team
xAI launched Team Bots: Grok Bots built around a role a whole team shares rather than one person's setup. Each one bundles the team's files, instructions and skills, plugins for apps such as Salesforce, Notion and GitHub, credentials for APIs that have no plugin, and memories it keeps as it learns the role. Everyone works with the same bot while their own conversations stay private, and a Team Bot can be invited into a Slack channel under its own handle.
- xAI
How SpaceXAI is using Grok Bot to scale customer support
xAI rebuilt its combined support operation around Grok Bot as Cursor's portfolio joined the company: ticket volume is up 175% with no new hires, and the bot signs into the same ticketing and issue-tracking tools the team uses. Resolution cost landed around $0.20 to $0.30, against the $1 to $4 per resolution typical of flat-rate AI support tools.
- xAI
Introducing Grok 4.7
xAI's new model for coding and knowledge work ships at the same $2 per million input and $6 per million output tokens as Grok 4.6. It was trained with a longer reinforcement-learning run weighted toward tasks that take hours, natively understands the Grok Bot harness, and leads Grok 4.6 on CursorBench 4.0, Terminal-Bench 4.0 and the Harvey legal benchmark.
- xAI
Introducing Grok Voice Transcribe 2.0
The new speech-to-text model is about twice as accurate as version 1.0 at the same price, built for flaky phone lines, competing voices and local accents rather than clean single-speaker audio. It ranks first for accuracy among 32 streaming models on Artificial Analysis, becomes the Speech-to-Text API default, and version 1.0 is deprecated in the coming weeks.
- xAI
Memory in Grok Build
Grok Build now records project conventions, decisions and facts in the background as you work, then reads them back before touching related code in a later session. Notes are stored as markdown topic files, with a /memory browser and a /dream command that consolidates recent notes.
- xAI
Biosecurity at the frontier
LatchBio's independent BioSecBench-Refusal evaluation found Grok 4.6 was the only model tested to score above 50% on both refusing disguised biohazards and completing routine biological work.
- xAI
Setting Grok Bot loose on procurement
xAI gave Grok Bot access to vendor spend, contracts, and usage data; it surfaced more than $100,000 in direct savings.
- xAI News
Designing Grok Bot for a world of persistent agents
xAI published a design write-up on the Grok Bot avatar system and how the product was shaped around agents that keep running between conversations.
- xAI News
Grok Bot for Enterprise
xAI made Grok Bot available to enterprises. Grok and Cursor Enterprise customers get two weeks of free usage and can invite their whole organization, including people without an existing seat.
- AdiiX on X
A 7-step operator's manual for multi-agent Grok Bot teams
A widely shared playbook (400K views) argues the single-bot era is over: build a Chief bot that routes and delegates, keep shared state in /workspace instead of chat, and gate every workflow with source, evidence, and action checks. Humans approve only money, publishing, and irreversible actions.
- Phosphen on X
The Grok Bot PMs on persistent chats and building your first bot
Two SpaceXAI product managers who ship Grok Bot walked through their own bot teams in a 47-minute interview, with the headline claim that you should never be watching a context meter. A companion guide covers bots, skills, routines, plugins, and the shared cloud computer.
- Elon Musk on X
Musk: automatic token optimization is coming to Grok Bot
Elon Musk confirmed that automatic token optimization is coming to Grok Bot, part of a broader efficiency push across xAI. No percentages or pricing details yet, but the feature would cut usage costs for bot teams at the platform level.
- Kingy AI
Grok Bot vs Grok Automations vs Grok Build: the builder's map
A comparison of xAI's three builder surfaces: autonomous bots, scheduled automations, and the free low-code Grok Build tool.
Found Grok news we missed? Send the link. Items are added with a summary and always link back to the source.