- The Rundown AI
- Posts
- Claude enters the reasoning era
Claude enters the reasoning era
PLUS: Qwen launches new open-source thinking model
Rowan Cheung
February 25, 2025
Good morning, AI enthusiasts. Reasoning models are the new battleground in AI—and Anthropic has officially entered the race with Claude 3.7 Sonnet, the world’s first-ever hybrid reasoning model.
After months of quiet, Anthropic is making a big splash on the AI leaderboards. But with DeepSeek’s aggressive open-source push and a unified GPT-5 on the horizon, how long will this dominance last?
In today’s AI rundown:
Claude 3.7 Sonnet with 'hybrid reasoning'
Qwen’s new open-source thinking model
How to talk to your apps with Concierge AI
AI agents get their own communication protocol
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
ANTHROPIC
Image source: Anthropic
The Rundown: Anthropic just released Claude 3.7 Sonnet, the world's first ‘hybrid reasoning’ AI that can combine instant responses with controllable extended thinking capabilities — alongside a new agentic coding tool called Claude Code.
The details:
Claude 3.7 Sonnet enables users to toggle between a standard and "extended thinking" mode, with the latter showing the AI’s reasoning via a scratchpad.
API users can precisely control how long Claude thinks (up to 128K tokens), allowing them to balance speed, cost, and quality based on task complexity.
The AI achieves SOTA performance on real-world coding benchmarks and agentic tool use, surpassing competitors like o1, o3-mini, and DeepSeek R1.
Anthropic also introduced Claude Code, a command-line coding agent that can edit files, read code, and write and run tests, in a limited research preview.
Why it matters: Anthropic has finally brought Claude into the reasoning era — with vastly improved coding benchmarks, precise thinking control, and a new agentic feature that points to a major push on the dev side. With OpenAI also planning a hybrid reasoner, 3.7 may be the start of the next leg up for a new generation of models.
TOGETHER WITH IMAGINE AI LIVE
The Rundown: IMAGINE AI LIVE '25 is a transformative three-day conference at the Fontainebleau Las Vegas where business leaders can connect with AI innovators and learn practical strategies for enterprise AI implementation.
Attend on May 28-30 and experience:
Enterprise AI strategies from world-class practitioners and thought leaders
Hands-on workshops covering practical applications of AI in business operations
Networking opportunities with hundreds of executives actively transforming their organizations with AI
Apply by Friday, February 28 with code IMAGINEAILIVE25 and save 25% on VIP and General Passes!
ALIBABA
Image source: Alibaba
The Rundown: Alibaba's Qwen team just released QwQ-Max-Preview, a new reasoning-focused AI that introduces thinking capabilities to their chat platform — while promising a full open-source release soon.
The details:
QwQ-Max-Preview is built on Qwen2.5-Max but significantly enhanced for deep reasoning, excelling in mathematics, coding, and agentic tasks.
The model introduces a "Thinking (QwQ)" feature to Qwen Chat that allows users to see the AI's reasoning process as it works through complex problems.
Qwen announced plans to open-source QwQ-Max and Qwen2.5-Max under an Apache 2.0 license soon, making the models freely available for developers.
The team will also release smaller variants like QwQ-32B for local deployment on devices with limited compute resources.
Why it matters: Reasoning has become the new competitive frontier in AI, and Qwen’s move to open-source their flagship reasoner could push the industry toward having these capabilities as a standard rather than a gated, premium feature. Open source is staying right on the heels of industry leaders — with Chinese labs leading the way.
AI TRAINING
The Rundown: Concierge is the first connected AI Assistant that can read and write to your favorite software tools, by finding and updating information across all your apps in real-time.
Step-by-step:
Create a Concierge account here to get started for free.
Click "Connect" and choose the tools you use — Gmail, Slack, Jira, Notion, HubSpot, or something else.
Type requests like "summarize my unread emails", “create a Jira ticket based on feedback in Slack" or “research these prospects from Salesforce.”
Watch as your AI assistant instantly finds information and takes action across your connected apps.
Pro tip: Start with one or two apps you use most frequently, then gradually add more as you get comfortable with the workflow.
PRESENTED BY INNOVATING WITH AI
The Rundown: Innovating with AI's new program, AI Consultancy Project, transforms AI enthusiasts into professional consultants — tapping into a market projected to reach $54.7B by 2032.
The 6-month program delivers:
Proven frameworks for client acquisition and service delivery
A step-by-step path to six-figure consulting income
Students who land their first AI client in as little as 3 days
AI & COMMUNICATION
Image source: Anton Pidkuiko on YouTube
The Rundown: Two developers just introduced Gibber Link, a sound-based communication protocol that allows AI agents to detect each other on calls and switch from human speech to direct data transmission — reducing time and compute costs.
The details:
Created by Anton Pidkuiko and Boris Starkov at ElevenLabs’ recent Hackathon, the project uses an open-source data-over-sound library called “ggwave.”
In the demo, an agent detects another AI on the phone and switches to dial-up-style ggwave audio signals with transcriptions, instead of normal voice.
Using the sound-level protocol instead of generating speech reduces compute costs by up to 90% and shortens communication time by as much as 80%.
The design also ensures clearer communication in noisy environments compared to traditional speech recognition-based systems.
Why it matters: AI voice agents are about to be everywhere, meaning the volume of AI-to-AI calls will grow exponentially (especially for businesses). This hackathon-winning project is a great look at how finding more efficient and cost-effective methods for these interactions might take AI communication down completely new paths.
QUICK HITS
🤖 Zapier Agents - Equip agents with internal data to work across 7,000+ apps
👀 SmolVLM2 - Small AI models to analyze videos on phones and laptops
🗣️ ElevenLabs Studio - Structure, edit, and generate long-form audio
🧬 Evo 2 - Biomolecular AI model for insights into DNA, RNA, and proteins
Luma Labs released a new Video to Audio feature in Dream Machine, enabling users to easily generate synced audio for video outputs.
Perplexity posted a teaser for a new agentic search browser called Comet, with a waitlist available to sign up for early access.
Salesforce and Google expanded their partnership to (among other things) integrate Gemini into Agentforce, allowing AI agents to process images, audio, and videos.
Alibaba revealed plans to invest $52B in cloud computing and AI infrastructure over the next three years, surpassing total spending in these sectors over the past decade.
Nothing unveiled its upcoming AI-powered Nothing Phone (3a), featuring an unboxing video by 1X’s recently debuted NEO Gamma humanoid robot.
Meta AI launched in Arabic across 10 Middle Eastern and North African countries, taking text generation, image creation, and animation to millions of new users.
COMMUNITY
Join our next workshop on Thursday, February 27, at 1 PM EST and learn how to transform your creative concepts into professional videos with LTX Studio's Tom Factor, Director of Creative Strategy and AI, and Alon Yaar, VP of Product.
RSVP here. Not a member yet? Join The Rundown University on a 14-day free trial.
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
That's it for today!Before you go we’d love to know what you thought of today's newsletter to help us improve The Rundown experience for you. |
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
Reply