@VentureBeat profile picture
VentureBeat @VentureBeat

Obsessed with covering transformative technology.

Joined July 2009
  • Tweets 180,904
  • Following 1,643
  • Followers 685,587
38,892 Photos and videos
Anthropic blocks all public access to Claude Fable 5, Mythos 5 following US government order — what enterprises should do venturebeat.com/ai/anthropic…

Anthropic blocks all public access to Claude Fable 5, Mythos 5 following US government order — what...

Enterprises can no longer afford — from an operational reliability standpoint — to run critical workflows on any single AI model or even provider.

venturebeat.com
1
4
5
1,751
Moonshot skipped DeepSWE. An independent KernelBench run found regressions. Practitioners building model routers are asking why. venturebeat.com/ai/open-sour…

Kimi K2.7-Code cuts tokens 30%, but skips independent benchmarks

Kimi K2.7-Code claims 30% fewer thinking tokens and a drop-in API swap path, but independent benchmarks show kernel regressions and no DeepSWE submission.

venturebeat.com
5
2
1,392
Google researchers introduce 'faithful uncertainty', allowing LLMs to offer best guesses instead of hallucinations venturebeat.com/ai/google-re…

Google researchers introduce 'faithful uncertainty', allowing LLMs to offer best guesses instead of...

Chasing zero hallucinations is costing you valid answers. Google researchers propose a "metacognitive" approach to save enterprise AI.

venturebeat.com
3
6
1,353
Speaker Reveal Friday: Animesh Singh, Sr. Director of AI Platforms & Infrastructure @LinkedIn, joins #VBTransform 2026. 250x GPU infrastructure growth. AI platforms for 1.3B members. Founder of Liger and KServe. Inference at scale, from someone who's done it. Join him this July 14-15 in Menlo Park: bit.ly/49oTA6r
1
1,172
Speaker Reveal Friday: Rachad Alao, VP of Product Engineering @cohere, joins #VBTransform 2026. He's building Cohere's Agentic AI platform, after leading AI Trust & Safety at Meta and Responsible AI at Google. Agentic autonomy, with guardrails. July 14–15, Menlo Park. Join us: bit.ly/49oTA6r
3
3
1,157
NanoClaw and JFrog launch 'immune system' to block AI agents from downloading malicious code venturebeat.com/ai/nanoclaw-…

NanoClaw and JFrog launch 'immune system' to block AI agents from downloading malicious code

This ensures that all agent activity adheres to the company’s specific commercial licenses, internal security policies, visibility needs, and governance standards.

venturebeat.com
3
8
1,191
Text parsers are destroying your RAG answers. UC Berkeley's PixelRAG skips them entirely — screenshots instead of parsed text — and beats text-based RAG on all six benchmarks. venturebeat.com/ai/pixelrag-…

PixelRAG beats text parsers, cuts agent costs 10x

UC Berkeley's PixelRAG renders pages as screenshots instead of parsing text, boosting RAG accuracy by up to 18.1% and cutting AI agent token costs 10x.

venturebeat.com
1
4
7
1,176
Something significant is happening in enterprise AI — and New York is at the center of it. VentureBeat's AI Impact Series returns to NYC on June 24 for Architecture of Velocity: a closed-door conversation about the multi-agent future and the architectural decisions that will define which enterprises lead and which fall behind. Fewer than 100 leaders. Invite-only. 13 days away. Request your invitation: bit.ly/4uCmNnx #AgenticAI #AIImpact #VentureBeat
1
953
Xiaomi's new open source, agentic AI coding harness MiMo Code beats Claude Code at ultra-long, 200 step tasks venturebeat.com/ai/xiaomis-n…

Xiaomi's new open source, agentic AI coding harness MiMo Code beats Claude Code at ultra-long, 200 ...

The persistent memory system addresses a real and widely felt pain point in agentic development workflows — one that competitors are also racing to solve.

venturebeat.com
1
6
4
1,485
Enterprise #AI teams face a dilemma: The best models today might not be the best models a year from now. MassMutual's answer is to stop making long-term bets — and build infrastructure that can swap models as the market shifts. #ArtificialIntelligence #GenerativeAI bit.ly/4e4nmRx

MassMutual AI: 12-month contracts, no lock-in

MassMutual caps vendor contracts at 12 months and runs a multi-model architecture — cutting contact center resolution times from 10 minutes to one.

venturebeat.com
2
1
5
1,174
New research cuts LLM context 16x before decode — 8.8x faster inference than every KV cache method tested. Open-sourced on HuggingFace. venturebeat.com/ai/context-c…

LLM context compression at 16x beats KV cache

LCLMs compress LLM context before decode — 8.8x faster at 16x compression, beating every KV cache method tested. Open-sourced by NYU and Columbia.

venturebeat.com
3
4
1,144
The #AI pilot worked. The demo landed. So why is it still not in production? Inside @CapitalOne's disciplined R&D approach to closing the gap between lab results and enterprise-scale AI. venturebeat.com/orchestratio… (Sponsored)
3
947
What AI benchmarks miss about real-world performance venturebeat.com/orchestratio…

What AI benchmarks miss about real-world performance

Enterprise AI teams have spent years solving for compute, securing GPU allocations, negotiating cloud capacity, and benchmarking training throughput. The assumption embedded in that work is that the...

venturebeat.com
795
Microsoft’s open-source SkillOpt automatically upgrades AI agent skills without touching model weights venturebeat.com/ai/microsoft…

Microsoft’s open-source SkillOpt automatically upgrades AI agent skills without touching model...

Microsoft's SkillOpt brings deep-learning discipline to AI agent skills, replacing manual prompt tweaking with mathematically validated text optimization.

venturebeat.com
2
5
955
The typewriter model is out. Google's DiffusionGemma generates text in parallel and self-corrects as it goes — 4x faster on a single GPU, with a quality trade-off it owns up to. venturebeat.com/ai/googles-d…

Google's DiffusionGemma runs text 4x faster

Google's open-source diffusion language model generates 256 tokens in parallel and self-corrects, hitting 4x speed on one GPU at a cost to quality.

venturebeat.com
4
9
1,182
Surprise upset: GPT-5.5 beats Claude Fable 5 on brutal new Agents’ Last Exam benchmark venturebeat.com/ai/surprise-…

Surprise upset: GPT-5.5 beats Claude Fable 5 on brutal new Agents’ Last Exam benchmark

The victory of GPT-5.5 aligns with recent third-party analysis suggesting that OpenAI's models are currently superior at strictly adhering to multi-part, complex prompts.

venturebeat.com
5
9
1,525
A 1B reasoning model trained on 40B tokens — not trillions — hit benchmark scores rivaling models 7x its size. Cost: $1,500. venturebeat.com/ai/researche…

A $1,500 foundation model that rivals larger LLMs

Sapient researchers trained a 1B reasoning model on just 40B tokens — scoring competitively with 2B-7B models at a fraction of typical pretraining cost.

venturebeat.com
3
4
1,200
Anthropic CEO calls for FAA-style regulation of powerful AI models: what enterprises should know venturebeat.com/ai/anthropic…

Anthropic CEO calls for FAA-style regulation of powerful AI models: what enterprises should know

To prepare for this shift, enterprises must first decouple their AI strategies from single-vendor dependencies. If a flagship model is suddenly blocked or recalled under the proposed FAA-style...

venturebeat.com
1
4
5
1,467
The VB Transform 2026 agenda is live. The pilot era is over. July 14–15 in Menlo Park, the engineers and architects scaling agentic AI in production take the stage. Visa. Target. Mastercard. Instacart. Brex. LangChain. And more. Real projects. Proprietary research. Zero vendor fluff. → bit.ly/49oTA6r
1
2
1
1,081
MassMutual's CIO replaced a 15-person SI team's 90-day mainframe analysis with a 7-day AI workflow — COBOL copy books, green screen screenshots, the whole thing. Sears Merritt joins Beyond the Pilot to break down that call — plus how they accidentally avoided the token cost explosion and got contact center resolution down from 10 minutes to 1. youtu.be/7IcU0QBrYng?si=Uu6_…

Weaponize Tokenmaxing: MassMutual’s ROI Engine

He negotiated seat-based (unlimited) licenses before token costs ex...

youtube.com
2
3
1,838
Load more