TESSFeed

Technical AI signals · Daily at 8 PM ET

Saturday, August 1, 2026

Proof Claims and Agent Runtime

Today split between OpenAI’s Lean-backed open-problem proof claim, DeepSeek turning long-context caching into a real latency win, and agent infrastructure spreading across memory, payments, and open runtimes.

🛰️  Top Signals

X / Twitter

Coding agents now handle video editing tasks

One list of claimed Codex abilities includes speeding up footage, trimming clips, converting MOV to MP4, extracting audio, and making vertical social-media versions.

@elder_plinius

Google is preparing Gemini 4

The post says Google has started pre-training a new foundation model, while Gemini 3.5 Pro stays in testing with trusted partners.

@Mr_Salio

Science paper reports a biomedical AI agent

The Science paper describes a general-purpose biomedical AI agent designed to automate biomedical research workflows, including basic research and translation.

@ScienceMagazine

Blogs

deepseek-ai/DeepSeek-V4-Flash-0731

Simon Willison says the 304B-parameter DeepSeek-V4-Flash-0731 costs $0.14 per million input tokens and $0.27 per million output tokens.

Simon Willison

datasette-apps 0.2a0

datasette-apps 0.2a0 adds app_debug(), which opens an app invisibly in a 0-opacity iframe and runs JavaScript to test it.

Simon Willison

Generalization and infinite width

The post summarizes a paper on the complexity of infinite-width networks and when functions can be learned with polynomial sample complexity.

LessWrong

The Epoch Brief - July 31, 2026

Epoch AI says the brief covers FrontierMath, parallelizability and technological singularity, AI energy use, and signs of AI uplift.

Epoch AI

Do your capabilities homework

The post argues technical AI safety researchers should pay more attention to RLVR trends, citing GRPO from the R1 paper 1.5 years ago.

LessWrong

Using AI to analyze life patterns

The post describes using AI to transcribe handwritten worksheets and diagrams from notes and reviews to extract patterns and forgotten insights.

LessWrong

Hacker News

Seedance 2.5

Commenters say Seedance 2.5 looks unusually good for video generation, with one top comment calling the washing-machine ad example as strong as social media content.

HN discussion

NetBSD 11.0

Commenters point to NetBSD 11.0 release details, including npf(7) layer 2 and user/group filtering and a new x86 MICROVM kernel that can boot in about 10 ms.

HN discussion

Linux on ESP32

Commenters say the repo claims a Linux-on-ESP32-31 setup that appears to boot enough to log in and run commands.

HN discussion

Research

YouTube

Hugging Face

deepseek-ai/DeepSeek-V4-Flash-0731

DeepSeek-V4-Flash-0731 is the official release of DeepSeek-V4-Flash and uses speculative decoding with the same model structure as DeepSeek-V4-Flash-DSpark.

model

moonshotai/Kimi-K3

Kimi K3 is a 2.8T-parameter open-weight multimodal model with a 1-million-token context window and native vision.

model

baidu/Unlimited-OCR

Unlimited OCR Works is Baidu's long-horizon parsing model, with vLLM inference support and a Hugging Face demo.

model

microsoft/Fara1.5-27B

Fara1.5-27B is a browser computer-use agent that acts from screenshots by emitting clicks, typing, scrolling, URL visits, and web searches.

model

zai-org/GLM-5.2

GLM-5.2 adds a stable 1M-token context and multiple thinking effort levels for coding.

model

upstage/Solar-Open2-250B

Solar Open 2 is a 250B-A15B MoE model that activates 15B parameters per token and targets tool calling and multi-step reasoning.

model

HuggingFaceCode/stack-v3-train

The Stack v3 is the largest open GitHub code dataset and includes full-repository context for pretraining code models.

dataset

nota-ai/Solar-Open2-250B-Nota-NVFP4

Nota AI released a 4-bit NVFP4 quantized version of Solar Open2 250B in llm-compressor format for vLLM, and the model requires a Blackwell GPU such as B200 or GB200.

model

owensong/Inflect-Micro-v2

Inflect-Micro-v2 is a sub-10M-parameter text-to-waveform speech model with fixed-voice English TTS, deterministic seeds, long-text handling, and CPU or CUDA inference.

model

GitHub

huggingface/speech-to-speech

This voice-agent pipeline combines VAD, STT, LLM, and TTS behind an OpenAI Realtime-compatible WebSocket API.

Python · ★ 10,180

microsoft/TRELLIS.2

TRELLIS.2 is a 4B-parameter image-to-3D model that uses an O-Voxel sparse voxel structure for high-fidelity assets.

Python · ★ 9,894

zhaoxuya520/reverse-skill

The package routes AI agents handling APKs, binaries, encrypted frontend JavaScript, CTFs, and pentesting targets to the right workflow.

PowerShell · ★ 11,822

github/gh-stack

gh-stack is a GitHub CLI extension for stacked branches and pull requests, and the repo also ships an AI agent skill for using it.

Go · ★ 797

usekaneo/kaneo

Kaneo is a project-management app built around fewer notifications, fewer buttons, and fewer workflows.

TypeScript · ★ 5,645

iv-org/invidious

Invidious is an open-source YouTube front end with no ads, no tracking, and no JavaScript required.

Crystal · ★ 21,591

microsoft/AI-For-Beginners

Microsoft's curriculum offers 12 weeks and 24 lessons covering AI tools, quizzes, labs, and ethics.

Jupyter Notebook · ★ 57,056

ansible/ansible

Ansible is agentless infrastructure automation built around SSH and is used for configuration management, deployment, cloud provisioning, and multi-node orchestration.

Python · ★ 70,089

Get this in your inbox

The same feed, delivered daily at 8 PM ET. No spam, unsubscribe anytime.