Category intelligence

Social Media Briefing — February 5, 2026

563 current items analyzed and ranked.

Executive synthesis

Social Media Summary

Super Bowl rivalry dominated AI discourse as Sam Altman fired back at Anthropic's ad campaign, defending OpenAI's free access model while announcing 500K Codex downloads since Monday—a rare public clash between frontier lab CEOs.

Andrej Karpathy marked the 1-year anniversary of "vibe coding" by proposing "agentic engineering" as the professional evolution, sparking discussion on AI-assisted programming paradigms
Google shared massive scale metrics: Gemini now processes 10B tokens/minute with 750M monthly active users
Mistral launched Voxtral 2 with open weights under Apache 2.0, intensifying speech-to-text competition
Shane Legg and François Chollet offered contrasting perspectives on AGI definitions, with Legg emphasizing that failing trivial human tasks disqualifies AGI claims

Allen AI's OpenScholar publication in Nature validated AI tools for scientific synthesis, while Perplexity unveiled upgraded Deep Research claiming SOTA benchmark performance. Anthropic added an /insights command to Claude Code for analyzing user history patterns.

Key Themes

OpenAI vs Anthropic Rivalry · 3Speech Recognition & Audio AI · 6Deep Research & Scientific AI · 10Deep Research Tools · 6Inference Optimization & GPU Infrastructure · 2Claude Code & Developer Tools · 5AGI Definition Debate · 5Infrastructure & Inference Optimization · 4AI Scale & Adoption Metrics · 2Codex Launch Success · 4

Primary evidence

Top Ranked Signals

Social Twitter Feb 4

First, the good part of the Anthropic ads: they are funny, and I laughed. But I wonder why Anthropi...

By @sama

98 score

AI Analysis

Sam Altman's extensive response to Anthropic's Super Bowl ad, defending OpenAI's free access model, announcing 500K Codex app downloads since Monday, criticizing Anthropic as 'authoritarian' for blocking competitors and controlling AI use cases. Major industry rivalry moment.

First, the good part of the Anthropic ads: they are funny, and I laughed. But I wonder why Anthropic would go for something so clearly dishonest. Our most important principle for ads says that we won’t do exactly this; we would obviously never run ads in the way Anthropic depicts them. We are not stupid and we know our users would reject that. I guess it’s on brand for Anthropic doublespeak to use a deceptive ad to critique theoretical deceptive ads that aren’t real, but a Super Bowl ad is not

OpenAI vs Anthropic rivalryAI business modelsCodex launch successAI democratization

Social Twitter Feb 4

A lot of people quote tweeted this as 1 year anniversary of vibe coding. Some retrospective - I've ...

By @karpathy

92 score

AI Analysis

Andrej Karpathy reflects on 1-year anniversary of coining 'vibe coding', proposes 'agentic engineering' as the professional evolution - emphasizing orchestrating agents with oversight while maintaining software quality. Notes 2026 will see improvements in both model and agent layers.

A lot of people quote tweeted this as 1 year anniversary of vibe coding. Some retrospective - I've had a Twitter account for 17 years now (omg) and I still can't predict my tweet engagement basically at all. This was a shower of thoughts throwaway tweet that I just fired off without thinking but somehow it minted a fitting name at the right moment for something that a lot of people were feeling at the same time, so here we are: vibe coding is now mentioned on my Wikipedia as a major memetic "co

Vibe coding evolutionAgentic engineeringAI-assisted developmentFuture of programming

Social Twitter Feb 4

Introducing Voxtral Transcribe 2, next-gen speech-to-text models by @MistralAI. State-of-the-art tra...

By @MistralAI

92 score

AI Analysis

Mistral AI announces Voxtral Transcribe 2, next-generation speech-to-text models with state-of-the-art transcription, speaker diarization, and sub-200ms real-time latency

Introducing Voxtral Transcribe 2, next-gen speech-to-text models by @MistralAI. State-of-the-art transcription, speaker diarization, sub-200ms real-time latency. Details in 🧵 t.co/0IeiJOpiAZ

product_launchspeech_recognitionopen_source

Social Twitter Feb 4

We've added a new command to Claude Code called /insights When you run it, Claude Code will read y...

By @trq212

92 score

AI Analysis

Anthropic announces new /insights command for Claude Code that analyzes a month of user message history to summarize projects, usage patterns, and provide workflow improvement suggestions

We've added a new command to Claude Code called /insights When you run it, Claude Code will read your message history from the past month. It'll summarize your projects, how you use Claude Code, and give suggestions on how to improve your workflow. t.co/xK7eN0qdB4

Claude Codedeveloper toolsAI agentsproduct launch

Social Twitter Feb 4

Gemini now processes over 10 billion tokens per minute via direct API use by our customers and the G...

By @OfficialLoganK

90 score

AI Analysis

Google's Logan K shares Gemini metrics: over 10 billion tokens processed per minute via API, and Gemini App has crossed 750 million monthly active users

Gemini now processes over 10 billion tokens per minute via direct API use by our customers and the Gemini App just crossed 750M monthly active users : )

GeminiGoogleAI scaleusage metrics

Social Twitter Feb 4

We've upgraded Deep Research in Perplexity. Perplexity Deep Research achieves state-of-the-art perf...

By @perplexity_ai

88 score

AI Analysis

Perplexity announces upgraded Deep Research achieving SOTA performance on benchmarks, outperforming competitors on accuracy and reliability

We've upgraded Deep Research in Perplexity. Perplexity Deep Research achieves state-of-the-art performance on leading external benchmarks, outperforming other deep research tools on accuracy and reliability. Available now for Max users. Rolling out to Pro in the coming days. t.co/8RAlewuWa3

product_launchdeep_researchbenchmarks

Social Twitter Feb 4

Our OpenScholar paper is now in @Nature 🎉 OpenScholar is an open-source model for synthesizing scie...

By @allen_ai

88 score

AI Analysis

Allen AI announces OpenScholar paper published in Nature - an open-source model for synthesizing scientific research with citations as accurate as human experts

Our OpenScholar paper is now in @Nature 🎉 OpenScholar is an open-source model for synthesizing scientific research—with citations as accurate as human experts. 🧵 t.co/vWVxlq3OQK

scientific AIcitation accuracyopen sourceresearch toolshallucination reduction

Social Twitter Feb 4

Today, we're rolling out an Advanced version of Perplexity Deep Research, achieving state-of-the-art...

By @AravSrinivas

88 score

AI Analysis

Perplexity CEO Arav Srinivas announces Advanced version of Deep Research, claiming state-of-the-art performance on benchmarks, beating competitors on accuracy, usability, and reliability

Today, we're rolling out an Advanced version of Perplexity Deep Research, achieving state-of-the-art performance on external and internal benchmarks, beating every other deep research tool on accuracy, usability, and reliability across all verticals. t.co/YkdqpX9I0k

product_launchesai_search_toolscompetitive_landscape

Social Twitter Feb 4

Physical AI is here. 🤝 Grounded in physics. Driven by intelligence. Together, @Dassault3DS and NV...

By @nvidia

87 score

AI Analysis

NVIDIA and Dassault3DS announce 'Physical AI' collaboration for virtual twins that learn, simulate and evolve - grounded in physics-based world models

Physical AI is here. 🤝 Grounded in physics. Driven by intelligence. Together, @Dassault3DS and NVIDIA are pioneering a new era of virtual twins that learn, simulate, and evolve. 🔗 t.co/SrTMcT6ndb t.co/AtTBHFN4Mz

physical_aipartnershipssimulation

Social Twitter Feb 4

Natural evolution suggests that AGI won't come from larger models that cram more and more specific k...

By @fchollet

85 score

AI Analysis

François Chollet argues AGI won't come from scaling up models with more knowledge, but from discovering meta-rules allowing systems to grow and adapt their own architecture in response to environment - drawing parallel to natural evolution.

Natural evolution suggests that AGI won't come from larger models that cram more and more specific knowledge, but from discovering the meta-rules that allow a system to grow and adapt its own architecture in response to the environment.

AGI architectureScaling debateAdaptive systems

Social Twitter Feb 4

🚀🚀🚀 vLLM on NVIDIA GB200: 26.2K prefill TPGS, 10.1K decode TPGS for DeepSeek R1/V3. 📈 3-5x throughpu...

By @vllm_project

85 score

AI Analysis

vLLM achieves 26.2K prefill TPGS on NVIDIA GB200 for DeepSeek R1/V3, showing 3-5x throughput vs H200 with half the GPUs through key optimizations

🚀🚀🚀 vLLM on NVIDIA GB200: 26.2K prefill TPGS, 10.1K decode TPGS for DeepSeek R1/V3. 📈 3-5x throughput vs H200 - with half the GPUs! Key optimizations:

NVFP4 GEMM for MoE experts
FP8 GEMM for MLA
Kernel fusion (RoPE+Quant+Q Write)
Weight offloading v2 with async prefetch

Thanks to the @AIatMeta and @NVIDIAAIDev teams for the collaboration! 🙏 🔗 Blog: t.co/CbdU85SGTd

infrastructurebenchmarksoptimization

Social Twitter Feb 4

📈 vLLM community + @nvidia pushed gpt-oss-120b performance on Blackwell GPUs to new heights: ⚡ +38%...

By @vllm_project

85 score

AI Analysis

vLLM project announces major performance improvements for gpt-oss-120b on NVIDIA Blackwell GPUs: +38% max throughput, +13% min latency. Achieved through FlashInfer integration, torch.compile kernel fusions, async scheduling. Collaboration between NVIDIA, RedHat, Meta, and vLLM community.

📈 vLLM community + @nvidia pushed gpt-oss-120b performance on Blackwell GPUs to new heights: ⚡ +38% max throughput 🎯 +13% min latency 📈 Entire Pareto frontier improved Key ingredients: FlashInfer integration, torch.compile kernel fusions, async scheduling, and stream interval optimizations. Deep dive + deployment recipes: t.co/Rdgeb317o0 Thanks to the teams at @NVIDIAAI , @RedHat_AI, @AIatMeta, and the vLLM community for the collaboration 🙏

inference_optimizationopen_source_aigpu_infrastructure