In partnership with
Hey AI Enthusiasts!
Yesterday was one of the biggest days in AI this year. Anthropic and OpenAI both dropped their most powerful coding models on the exact same day and they're not even the only story. OpenAI also launched a full enterprise agent platform and Perplexity quietly introduced a feature that could change how we trust AI answers.
Let’s dive in!
In today’s insights:
🚨 Anthropic launches Claude Opus 4.6 with "agent teams"
🤖 OpenAI launches GPT-5.3-Codex (self-developing AI) + Frontier platform
🧠 Perplexity just dropped Model Council
Read time: 5 minutes.
🗞️ Recent Updates
The AI Field: Anthropic just released Claude Opus 4.6 their most powerful model ever, featuring a new "agent teams" capability that lets multiple AI agents split tasks and work in parallel.
Details:
Scores 65.4% on Terminal-Bench (up from 59.8%) and 72.7% on OSWorld, putting it ahead of GPT-5.2 and Gemini 3 Pro
New "agent teams" feature splits development work across coordinating AI agents that work in parallel autonomously
1 million token context window now available in beta, a first for Opus-class models
New "adaptive thinking" lets the model adjust how much reasoning it applies based on task complexity
Anthropic tested it by having agent teams build a full C compiler over 2,000 sessions and 2 billion tokens, costing just under $20,000
Found 500 zero-day vulnerabilities in open-source code, outperforming previous models 38 out of 40 times in blind cybersecurity tests
Why This Matters: The agent teams feature is the headline here. Instead of one AI working on a task, Opus 4.6 can split work across multiple agents that coordinate with each other, think code reviews, large codebase analysis, or any task where parallel processing makes sense. The C compiler demo is a flex: Anthropic essentially let a team of Claudes build a working compiler with minimal human intervention. The 1M token context window also means entire codebases can fit in a single conversation. This is Anthropic's strongest enterprise play yet.
Unlock ChatGPT’s Full Power at Work
ChatGPT is transforming productivity, but most teams miss its true potential. Subscribe to Mindstream for free and access 5 expert-built resources packed with prompts, workflows, and practical strategies for 2025.
Whether you're crafting content, managing projects, or automating work, this kit helps you save time and get better results every week.
The AI Field: OpenAI made a double move yesterday, releasing GPT-5.3-Codex (their most capable coding model) and launching Frontier, a brand new platform for enterprises to build and manage AI agents.
GPT-5.3-Codex Details:
Sets new industry highs on SWE-Bench Pro and Terminal-Bench for agentic coding
First OpenAI model that was instrumental in creating itself—the Codex team used early versions to debug its own training and diagnose its own evaluations
25% faster than GPT-5.2-Codex while using fewer tokens
Interactive while working—you can steer and redirect it mid-task without losing context
Rated "High" on OpenAI's Cybersecurity Preparedness Framework—the first model to receive this classification—meaning API access is being delayed as a precaution
OpenAI Frontier Details:
New platform for enterprises to build, deploy, and manage AI agents across their organization
Works with agents from any provider, not just OpenAI—an open standards approach
Connects siloed systems (CRM, data warehouses, ticketing tools) to give AI agents shared business context
Agents build memory over time, improving performance through past interactions
HP, Intuit, Oracle, State Farm, Thermo Fisher, and Uber are early adopters
Directly competes with Microsoft's Agent 365 and Anthropic's Claude Cowork
Why This Matters: Two stories, one theme: OpenAI is going all-in on the enterprise. GPT-5.3-Codex is their coding weapon—a model that literally helped build itself—while Frontier is the infrastructure play to get AI agents deployed inside large organizations. The cybersecurity flag on Codex is worth watching. OpenAI is essentially saying this model is so capable at code that it poses new risks. The Frontier launch, with its open-standards approach, is a smart move to become the "operating system" for enterprise AI—regardless of which models companies use.
The AI Field: Perplexity just launched Model Council, a feature that runs your query across three AI models simultaneously, then synthesizes the results into a single answer that shows where they agree and where they differ.
Details:
Runs queries across models like Claude Opus 4.6, GPT-5.2, and Gemini 3.0 at the same time
A synthesizer model reviews all outputs, resolves conflicts, and delivers one unified answer
Shows where models agree and where they disagree—so you can see the blind spots
Best for investment research, complex decisions, creative brainstorming, and fact verification
Available now for Perplexity Max subscribers ($200/month or $2,000/year) and Enterprise Max
Why This Matters: Every AI model has blind spots. Claude might miss something GPT catches, and vice versa. Model Council essentially eliminates single-model risk by cross-referencing multiple models before giving you an answer. For anyone making decisions based on AI research—investors, strategists, founders—this is a big deal. It's the difference between asking one expert and polling a panel. The $200/month price tag is steep for casual users, but for professionals making high-stakes decisions, it could easily pay for itself.
Learn how to make AI work for you
AI won’t take your job, but a person using AI might. That’s why 2,000,000+ professionals read The Rundown AI – the free newsletter that keeps you updated on the latest AI news and teaches you how to use it in just 5 minutes a day.
🛠️ Trending AI Tools
🎬 Deevid.ai – Create videos from text, images, or existing clips with AI-powered templates and effects, perfect for fast content creation and social posts.
🤖 CustomGPT - Build custom AI chatbots trained on your business content with no coding required.
🎙️ ElevenLabs - Generate ultra-realistic AI voices, clone your own voice, and dub videos in 70+ languages.
⚡ Base44 - Turn ideas into fully functional web apps in minutes using natural language prompts.
🎨 Adcreative - Generate high-converting ad creatives, banners, and videos using AI for any platform.
📧 Plusvibe - AI-powered cold email automation with inbox warm-up, lead enrichment, and deliverability optimization.
✍️ writesonic - AI writing and SEO platform for articles, ad copy, and search-optimized content.
🚀 Emergent - Build full-stack web and mobile apps in minutes using natural language prompts, backed by Y Combinator.
✨ Gamma - Create stunning presentations, websites, and documents with AI-powered design—no design skills needed.
🗞️ More AI Hits
💰 Amazon plans to spend $200 billion on AI in 2026 — stock drops 6% Amazon just announced the biggest AI spending plan of any tech company. The $200B capex forecast blew past Wall Street's $150B estimate. Most of it goes to AWS data centers and AI chips. CEO Andy Jassy says new AI capacity is being monetized "immediately," but investors aren't convinced yet.
📉 Anthropic's Claude Cowork triggers $285B software stock wipeout New Cowork plugins for legal, finance, and marketing sent shockwaves through Wall Street. Thomson Reuters dropped 16%, LegalZoom fell 20%, and the entire SaaS sector took a hit. Investors are now asking: if AI can do the work, why pay for the software? The selloff marks a shift from "AI helps software" to "AI replaces software."
🔵 Alphabet commits up to $185B in AI spending for 2026
Google's parent company nearly doubled its capex plans, committing $175–185 billion to AI infrastructure this year. Unlike Amazon, investors gave Alphabet a pass — strong cloud revenue growth backed up the spending. The AI infrastructure race is now a three-way battle between Alphabet, Amazon, and Meta.
🏥 Goodfire raises $150M to decode how AI models actually think
AI infrastructure startup Goodfire just closed a $150M Series B at a $1.25B valuation. The company builds tools that let enterprises inspect and fix how foundation models make decisions — catching bias, hallucinations, and failure modes before they cause damage. B Capital led the round with Menlo Ventures and Lightspeed joining.
WRAP-UP
That's all for todays roundup!
Anthropic drops Opus 4.6 with agent teams and a 1M token context window
OpenAI fires back with GPT-5.3-Codex (self-developing AI) and Frontier (enterprise agent platform)
Perplexity launches Model Council—multi-model answers that show where AIs agree and disagree
February 5, 2026 might be the biggest single day in AI this year
What did you think of today’s edition?
Until next time!
Olle | Founder of The AI Field