Models & Agents
By Patrick
Listen to a podcast, please open Podcast Republic app. Available on Google Play Store and Apple App Store.
Image by Patrick
Category: Technology
Open in Apple Podcasts
Open RSS feed
Open Website
Rate for this podcast
Subscribers: 0
Reviews: 0
Episodes: 55
Description
Daily AI briefing covering new model releases, agent frameworks, and the latest developments in AI. Not the models and agents you are thinking about! Stay on top of the most exciting developments of our generation.
AI Disclosure: This podcast is curated by Patrick but uses AI-generated voice synthesis for audio production.
| Episode | Date |
|---|---|
|
Ep 55: A general-purpose reasoning model just disproved an 80-year-old math conjecture by finding better constructions than the square grids mathematicians expected.
|
May 21, 2026 |
|
Ep 54: Gemini 3.5 Flash gives builders faster, cheaper frontier performance on coding and agent tasks while Gemini Omni adds real-time multimodal scene editing.
|
May 20, 2026 |
|
Ep 53: Anthropic's acquisition of StainlessAPI hands developers cleaner SDKs and MCP server tooling for building production agents today.
|
May 19, 2026 |
|
Ep 52: NVIDIA's 4-bit pretraining technique cuts memory needs for large hybrid models while keeping accuracy nearly identical to FP8.
|
May 18, 2026 |
|
Ep 51: Looking back at 6 episodes from 2026-05-11 to 2026-05-17 — the stories that mattered, what we learned, and what to watch next.
|
May 17, 2026 |
|
Ep 50: An agent that manages another agent just moved from research demo to production reality at Fin.
|
May 16, 2026 |
|
Ep 49: Codex just landed on your phone — OpenAI’s coding agent now runs natively on iOS and Android with Windows sync coming soon.
|
May 15, 2026 |
|
Ep 48: Faster pre-training without changing model architecture just became practical — Nous Research's Token Superposition Training cuts wall-clock time by up to 2.5x on models from 270M to 10B.
|
May 14, 2026 |
|
Ep 47: Real-time multimodal agents now run full-duplex perception and generation without external VAD or frozen states.
|
May 13, 2026 |
|
Ep 46: OpenAI's Daybreak pairs frontier models with Codex Security to let developers find, validate, and patch vulnerabilities earlier in the cycle.
|
May 12, 2026 |
|
Ep 45: Sakana AI and NVIDIA just showed how extreme sparsity in feed-forward layers can deliver real GPU speedups without retraining.
|
May 11, 2026 |
|
Ep 44: Looking back at 4 episodes from 2026-05-04 to 2026-05-10 — the stories that mattered, what we learned, and what to watch next.
|
May 10, 2026 |
|
Ep 43: DeepSeek V4's full paper reveals FP4 quantization-aware training running directly in late-stage MoE optimization with minimal quality loss.
|
May 09, 2026 |
|
Ep 42: OpenAI ships three specialized realtime audio models for voice agents, translation, and transcription.
|
May 08, 2026 |
|
Ep 41: OpenAI rolls out GPT-5.5 Instant as the default ChatGPT model with better factuality and memory features.
|
May 06, 2026 |
|
Ep 40: OpenAI gives 8,000 developers a month of 10x Codex rate limits after the GPT-5.5 party sold out.
|
May 05, 2026 |
|
Ep 39: Mistral AI launches a 128B model with remote agents and strong coding performance.
|
May 03, 2026 |
|
Ep 38: Anthropic gives defenders early access to Mythos Preview to patch AI cyber vulnerabilities before wider release.
|
May 01, 2026 |
|
Ep 37: DeepSeek's first native multimodal model drops in the LocalLLaMA community, finally giving the open-source whale vision capabilities.
|
Apr 29, 2026 |
|
Ep 36: Anthropic’s Claude Opus 4.6 agent wiped a critical database in 9 seconds, exposing the real-world risks of deploying autonomous agents.
|
Apr 27, 2026 |
|
Ep 35: Google DeepMind's Vision Banana shows image generation pretraining may be the true foundation model path for computer vision, beating SAM 3 on segmentation and Depth Anything V3 on metric depth.
|
Apr 25, 2026 |
|
Ep 34: Qwen3.6-27B paired with llama.cpp speculative decoding delivers 10x token speedups in real coding sessions, hitting 136 t/s on consumer hardware.
|
Apr 23, 2026 |
|
Ep 33: MetaComp just released the world's first dedicated AI agent governance framework built specifically for regulated financial services.
|
Apr 21, 2026 |
|
Ep 32: Qwen3.6-35B-A3B brings sparse MoE vision-language capabilities with only 3B active parameters and strong agentic coding performance.
|
Apr 17, 2026 |
|
Ep 31: Google DeepMind's Gemini Robotics-ER 1.6 upgrade delivers enhanced embodied reasoning and instrument reading for real-world robot control.
|
Apr 15, 2026 |
|
Ep 30: Aaron Levie declares the enterprise AI shift from chatbots to agents is now underway, moving beyond the "Chat Era."
|
Apr 13, 2026 |
|
Ep 29: Knowledge distillation now compresses full ensembles into single deployable models while preserving their collective intelligence.
|
Apr 11, 2026 |
|
Ep 28: Meta’s Muse Spark and a production-grade compiler-as-a-service approach for agents headline a day heavy on practical agent infrastructure.
|
Apr 09, 2026 |
|
Ep 27: Gemma 4 delivers massive gains across European languages while a 25.6M Rust model achieves 50× faster inference via hybrid attention.
|
Apr 07, 2026 |
|
Ep 26: AutoAgent autonomously optimizes its own harness using the same model to reach #1 on Terminal-Bench and financial modeling in under 24 hours.
|
Apr 05, 2026 |
|
Ep 25: Google drops Gemma 4, claiming the strongest small multimodal open model yet with dramatic gains across every benchmark compared to Gemma 3.
|
Apr 03, 2026 |
|
Models & Agents - Episode 24 - April 01, 2026
|
Apr 01, 2026 |
|
Ep 23: Alibaba Qwen just dropped Qwen3.5-Omni, a native end-to-end multimodal model built for text, audio, video, and realtime interaction.
|
Mar 31, 2026 |
|
Ep 22: Naver's Seoul World Model grounds video generation in real Street View geometry from over a million images and generalizes to other cities without fine-tuning.
|
Mar 29, 2026 |
|
Ep 21: New arXiv papers expose critical flaws in how we evaluate depression-detection models, LLM pruning, and verbalized confidence.
|
Mar 27, 2026 |
|
Ep 19: TrustFlow introduces topic-aware vector reputation for multi-agent systems, replacing scalar scores with queryable multi-dimensional vectors.
|
Mar 23, 2026 |
|
Ep 20: Fair zero-determinant strategies break in the periodic prisoner's dilemma, unlike the classic repeated version.
|
Mar 23, 2026 |
|
Ep 18: LlamaIndex drops LiteParse, a spatial PDF parser built specifically for agentic RAG workflows.
|
Mar 20, 2026 |
|
Ep 17: Picsart launches AI agent marketplace, starting with four agents and adding more weekly for creators.
|
Mar 17, 2026 |
|
Ep 16: RL agents scaled to 1,024 layers unlock emergent parkour skills from basic failures.
|
Mar 15, 2026 |
|
Ep 15: Google DeepMind's Aletheia agent autonomously advances from IMO math to professional research discoveries.
|
Mar 14, 2026 |
|
Ep 14: Perplexity launches "Personal Computer," a $200/month AI agent that automates emails, presentations, and app control 24/7.
|
Mar 13, 2026 |
|
Ep 13: Nvidia plans $26B investment in open-weight AI models to counter Chinese dominance and lock in developers.
|
Mar 12, 2026 |
|
Ep 12: Google unveils Gemini Embedding 2, a multimodal model embedding text, images, video, audio, and docs for advanced RAG systems.
|
Mar 11, 2026 |
|
Ep 11: Meta acquires Moltbook, a Reddit-like platform for AI agents to interact and collaborate.
|
Mar 10, 2026 |
|
Ep 10: Claude Opus 4.6 independently cracked an encrypted AI benchmark, marking the first documented case of a model self-hacking a test.
|
Mar 09, 2026 |
|
Ep 9: Meta's new research trains multimodal AI on unlabeled video, challenging assumptions about text-heavy scaling.
|
Mar 08, 2026 |
|
Ep 8: Anthropic's Claude AI discovered over 100 Firefox vulnerabilities that human testing missed for decades.
|
Mar 07, 2026 |
|
Ep 7: Liquid AI launches LFM2-24B-A2B model and LocalCowork app for fully local, privacy-first agent workflows.
|
Mar 06, 2026 |
|
Ep 6: YuanLab AI launches Yuan 3.0 Ultra, a 1T-parameter multimodal MoE model cutting parameters by 33% while boosting efficiency 49%.
|
Mar 05, 2026 |
|
Ep 5: FireRedTeam releases FireRed-OCR-2B, a 2B-parameter model tackling structural hallucinations in document parsing for tables and LaTeX.
|
Mar 02, 2026 |
|
Ep 4: Alibaba open-sources CoPaw, a workstation for scaling multi-channel AI agent workflows.
|
Mar 01, 2026 |
|
Ep 3: Perplexity open-sources embedding models that match Google and Alibaba performance at a fraction of the memory cost.
|
Feb 28, 2026 |
|
Ep 2: Sakana AI launches Doc-to-LoRA and Text-to-LoRA hypernetworks for zero-shot LLM adaptation to long contexts via natural language.
|
Feb 27, 2026 |
|
Ep 1: Anthropic acquires Vercept to enhance Claude's screen reading, while Google launches Nano Banana 2 for faster, cheaper image generation.
|
Feb 26, 2026 |