Skip to main content
Back

New DeepSeek V3.1: Shifting from Chatbots to Agents

On DeepSeek, Jupyter Agent 2, ElevenLabs new Alpha, Copilot, Claude Code, and more..

August 21, 2025 4 min read

Hey

Welcome to AlphaSignal, the most read source of news by AI engineers and researchers. Every day, we identity and summarize the top 1% of news, papers, models, and repos, so you’re always up to date.


Here’s today’s roundup:

./SUMMARY

Read time: 4 min 23 sec

Top News

▪ New DeepSeek V3.1: Shifting from Chatbots to Agents

Trending Signals

▪ Anthropic adds Claude Code to Enterprise and Team plan

▪ Hugging Face releases Jupyter Agent 2 for in-notebook coding and analysis

▪ ElevenLabs launches v3 (alpha) API for expressive text-to-speech

▪ NVIDIA unveils tiny open-source reasoning models, 6× faster than rivals

▪ GitHub integrates Copilot coding agent with GitHub Mobile and VS Code

Top Tutorials

▪ Google shows how to create Market Research Agents with Gemini

▪ Build an MCP workflow for image generation with Claude

▪ GitHubs tutorial on using Copilot to improve code reviews

Top Read

▪ Google’s guide to choosing the right AI developer tool

If you're enjoying AlphaSignal please forward this email to a colleague. 

It helps us keep this content free.

./TOP_NEWS

DeepSeek Releases V3.1: Shifting From Chatbots to Agents

13,385 Likes

alpha_signal_image_2

DeepSeek released V3.1 today, calling it its first major step into the agent era. Unlike earlier versions focused on raw reasoning benchmarks, V3.1 is designed to power autonomous workflows, enabling agents that can plan, call tools, and execute long multi-step tasks. 


The launch reflects a broader industry move away from static chatbots toward general-purpose reasoning systems built to act, not just answer.


What’s New in V3.1

  • Hybrid inference: one model supports both thinking (deep reasoning) and non-thinking (fast response) modes.

  • Agent upgrades: more reliable tool use with strict JSON, better multi-step reasoning, stronger performance in code automation and search agents.

  • Efficiency: comparable accuracy to R1-0528 but faster in reasoning mode.

  • 128K context: extended for large documents, workflows, and multi-agent pipelines.

  • Benchmarks: SWE-bench Verified 66.0 vs. 44.6 (R1); Terminal-Bench 31.3 vs. 5.7 (R1); AIME 2025 ~88%.

Technical Details

  • Size: 671B parameters (37B active per token).

  • Training: +840B tokens for context extension.

  • Compatibility: Anthropic API format, strict function calling, updated tokenizer and chat template required.

Pricing (Effective Sep 5, 2025, 16:00 UTC)

  • Input: $0.07 / 1M (cache hit), $0.56 / 1M (cache miss).

  • Output: $1.68 / 1M.

Availability

  • APIs:

    • deepseek-chat (non-thinking).

    • deepseek-reasoner (thinking).

  • Both support 128K context.

  • Available in DeepSeek Chat with a “DeepThink” toggle.

  • Open-source weights on Hugging Face and ModelScope.

Try now

./TOP_SIGNALS

Anthropic lets Enterprise and Team admins assign standard or premium seats with Claude Code access

2,586 Likes

Hugging Face introduces Jupyter Agent 2, running code and plots directly inside notebooks in real time

1,947 Likes

ElevenLabs debuts v3 (alpha) API featuring dialogue mode for lifelike multi-speaker conversations

1,395 Likes

NVIDIA ships open-source Nemotron Nano 2 (9B–12B), offering 6× faster reasoning than similar-sized models

1,048 Likes

GitHub rolls out Agents Panel, letting you launch Copilot coding tasks from any page on GitHub

659 Likes

./TOP_TUTORIALS

Google shows how to create Market Research Agents with Gemini

642 Likes

This tutorial shows how to build a TypeScript Node.js app using the Vercel AI SDK and Gemini API. It fetches market trends via Google Search, extracts structured data to generate charts, compiles findings into an HTML report, and exports it as a PDF.

Build an MCP workflow for image generation with Claude

603 Likes

This guide shows how to integrate Claude with Hugging Face Spaces through MCP. You build an MCP server, integrate FLUX.1 Krea for generation, and let Claude handle prompts and iteration to create structured, tool-enabled image workflows.

GitHub’s tutorial on using Copilot to improve code reviews

569 Likes

GitHub argues that faster coding via AI only matters when paired with expertise, i.e., knowing why code works, catching edge cases, and documenting clearly. It calls out pull requests, code reviews, and docs as foundational areas to master to thrive in AI-powered workflows.

./TOP_READ

Google’s guide to choosing the right AI developer tool

⇧ 293 Likes

alpha_signal_image_3

Google published a guide to help you choose the right AI tool for your workflow. Instead of covering everything, it gives clear recommendations depending on your goals and where you work.


The guide is structured as a decision aid, helping you match workflows with tools.


Key takeaways:

  • Gemini app: Use for quick experiments, prototyping, or casual exploration.

  • Gemini Code Assist: Use for AI-assisted coding inside IDEs.

  • Firebase Studio: Use if you build mobile or web apps and want integrated AI workflows.

  • Gemini CLI: Use for command-line workflows, automation, and flexible scripting.

  • Jules: Use if you need an AI teammate for collaboration.

This article helps you quickly map tasks to the right Google AI tool so you don’t waste time testing everything.

Read now

At Alpha Signal, our mission is to build a sharp, engaged community focused on AI, machine learning, and cutting-edge language models, helping over 200,000 developers stay informed and ahead. We’re passionate about curating the best in AI, from top research and trending technical blogs to expert insights and tailored job opportunities. We keep you connected to the breakthroughs and discussions that matter, so you can stay in the loop without endless searching. We also work closely with partners who value the future of AI, including employers and advertisers who want to reach an audience as passionate about AI as we are.


Our partnerships are based on shared values of ethics, responsibility, and a commitment to building a better world through technology.Privacy is a priority at Alpha Signal. Our Privacy Policy clearly explains how we collect, store, and use your personal and non-personal information. By using our website, you accept these terms, which you can review on our website. This policy applies across all Alpha Signal pages, outlining your rights and how to contact us if you want to adjust the use of your information. We’re based in the United States. By using our site, you agree to be governed by U.S. laws.

Looking to promote your company, product, service, or event to 200,000+ AI developers? Let's work together.

Work with us