About
All Posts
Partner
Logo
Login
Subscribe
Logo
  • Home
  • Posts
  • Kimi K3 debuts as world’s largest open AI model from China, closing in on U.S. rivals

Jul 17, 2026

Kimi K3 debuts as world’s largest open AI model from China, closing in on U.S. rivals

Bay Area Times
Bay Area Times

Friday 17 July 2026

Top stories today:

  1. Kimi K3 debuts as world’s largest open AI model from China

  2. Alphabet -4.4% as Gemini 3.5 Pro launch said to be delayed by months

  3. Netflix Q2 revenue +13% YoY, to end regular engagement updates

  4. Microsoft to unveil AI security tool as lower-cost Mythos alternative

  5. Databricks to raise $3B at $188B valuation in Coatue-led round

Follow us on X | Instagram | Advertise

0. Data & Calendar

All values as of 6 AM ET / 3 AM PT, other than S&P500 and NASDAQ close (4 PM ET / 1 PM PT).

All times are ET.

Listen to our AI-generated podcast summarizing today’s newsletter (beware of hallucinations):

1. Kimi K3 debuts as world’s largest open AI model from China, closing in on U.S. rivals

Kimi K3 scored 57 on the Artificial Analysis Intelligence Index, comparable to OpenAI’s GPT-5.5 and Claude Opus 4.8:

X: Artificial Analysis.

  • 2.8T total parameters in size.

  • Jul. 27: Kimi K3’s weights will be released by Moonshot AI.

  • Costs $3/$15 per 1M input/output tokens, versus GPT-5.5’s $5/$30, Claude Opus 4.8’s $5/$25, and GLM 5.2’s $1.4/$4.4.

  • Surpassed Claude Fable 5 on the Frontend Code Arena and only behind GPT-5.6 Sol on Terminal Bench 2.1.

Kimi K3’s hallucination rate has regressed compared with K2.6

X: Artificial Analysis.

2. Smart hiring for teams that need strong talent now*

Athyna.

  • Fill key roles faster, without waiting weeks or months for the pipeline to behave.

  • AI-assisted matching and structured vetting help you find people who fit the role and the team.

  • Cut time and cost with a process built for scale, not endless back-and-forth.

Build your team

*Sponsored.

3. Alphabet -4.4% as Gemini 3.5 Pro launch said to be delayed by months

Google Finance: Google stock.

  • Google is reportedly trying to improve Gemini 3.5 Pro’s capabilities, particularly in coding.

  • Engineers, AI researchers, and managers have grown frustrated with the delays.

  • Google has also been in talks with the U.S. government about the model’s capabilities and safety standards.

  • “We’re currently testing 3.5 Pro, an upgraded Flash model, and other models with partners,” the company said.

  • May: Gemini 3.5 Pro was widely expected to debut at Google I/O.

4. The HR-IT guide every team needs*

HR and IT need to work as one. Here's how

When HR and IT don't talk, people fall through the cracks. This guide fixes the handoffs that matter most.

*Sponsored.

5. Netflix Q2 revenue +13% YoY to $12.6B, to end regular engagement updates in 2027

  • View hours +2% in H1 2026, versus +1.5% growth in 2025.

  • +12% in Q3 revenue expected.

  • FY revenue projected at $51B-$51.4B, vs. prior guidance of $50.7B-$51.7B.

  • ~300 titles used GenAI this year, mostly in post-production, “to deliver higher quality output more quickly and at a lower cost,” Netflix said.

  • Shares -8% in after-hours trading as investors appeared disappointed by the earnings outlook.

6. Goodies delivered straight into your inbox*

Open Source CEO.

Get the chance to peek inside founders and leaders’ brains and see how they think about going from zero to 1 and beyond.

Join thousands of weekly readers at Google, OpenAI, Stripe, TikTok, Sequoia, and more.

Check all the tools and more here, and outperform the competition.

See what it's all about

*Sponsored.

7. Microsoft said to unveil AI security tool as lower-cost Anthropic’s Mythos alternative

  • Codenamed Project Perception, the product would use AI models from Anthropic, OpenAI and Microsoft to identify software bugs.

  • This month: the tool is expected to debut.

  • A model router would switch between models based on the task, helping keep costs below those of relying on a single model.

  • Relatedly, Satya Nadella criticized Claude Fable 5 internally as “editorially controlled,” saying its refusals “make no sense.”

Subscribe to keep reading

This content is free, but you must be subscribed to Bay Area Times to continue reading.

Already a subscriber?Sign in.Not now

Keep reading

Jul 16, 2026

Thinking Machines debuts Inkling, top U.S. open-weight AI model

Thinking Machines debuts Inkling, top U.S. open-weight AI model

Jul 15, 2026

OpenAI’s first device said to be movable, screen-free AI companion speaker

OpenAI’s first device said to be movable, screen-free AI companion speaker

Jul 14, 2026

OpenAI said to still target 2027 device launch despite Apple lawsuit

OpenAI said to still target 2027 device launch despite Apple lawsuit

Jul 13, 2026

Apple sues OpenAI over alleged trade secret theft, threatening hardware ambitions

Apple sues OpenAI over alleged trade secret theft, threatening hardware ambitions

Subscribe

Your morning briefing, free forever.

Get the latest news, resources, and more delivered straight to your inbox everyday.

The visual daily newsletter on business and tech. Analyzing the news with 1 visual per story. Join our 250,000+ subscribers here.

Copyright © 2026 Bay Area Times.

BAT

Company

Legal

Privacy Policy

Terms of Service