Skip to content
DigIntoAI/Model intelligence
Model intelligence feed

Model intelligence.

Follow the changes. Understand what comes next.

Editor’s pickOpenAI
OpenAI/GPT-6.2 Astra / Sol

OpenAI Launches GPT-6.2 Astra and GPT-6.2 Sol Across ChatGPT Work, Codex, and the API

At 01:45:00 Asia/Shanghai on November 18, 2026 (OpenAI News RSS; confirmed on @OpenAI at 01:54:03 Asia/Shanghai), OpenAI launched flagship GPT-6.2 Astra (gpt-6.2-astra) and everyday GPT-6.2 Sol (gpt-6.2-sol) on ChatGPT Work, Codex, and the API, bringing Ultrafast to Pro 200 starting with Sol, plus the gpt-6.2-astra-law configuration.

Latest stories485

Newest published
349

1Password for Agents Lets AI Agents Sign In Without Seeing Your Password

1Password for Agents lets a personal AI agent request a vault login; you approve, and 1Password fills the site so the agent never sees the password.

1Password for AgentsRead story
350

Uber Rolls Out ChatGPT Work to More Than 30,000 Employees

Uber expanded ChatGPT Work from a 1,000-person pilot to more than 30,000 employees. Default is GPT-6.1 Sol; engineering still uses Codex.

ChatGPT WorkRead story
351

METR and Redwood Red-Teamed OpenAI's Internal Agent Monitoring Before July 21

METR and Redwood spent three weeks inside OpenAI testing agent monitoring as of July 21. OpenAI says production gaps are patched; METR has not rechecked.

Internal agent monitoringRead story
352

iFlytek Releases Spark X2.6 with a 1M Context Window and Limited-Time API Prices

iFlytek's Spark X2.6 is a 293B-A30B MoE model with a 1M context window, aimed at code and agents. iFlytek says it was trained on domestic compute.

iFlytek Spark X2.6Read story
353

Hugging Face Says Its Open DNA Language Model Carbon-X Nears GPT-6.1 Sol at About 1/1,800 the Cost

Hugging Face's Carbon-X is a 14.8B open DNA language model built on Qwen3-14B, Apache 2.0. It claims near GPT-6.1 Sol scores at about 1/1,800 the cost.

Carbon-XRead story
354

Phonon Local Speech with Ollama: Speak, Export WAV, and Use the Speech API

Gradium released Phonon as open-weight text-to-speech on October 20. The smallest pack, Phonon 31m, is English-only under Apache 2.0. Besides Gradium's Python and browser paths, Ollama on Apple Silicon (v0.40.9 or later) can pull phonon for terminal speak, WAV export, and /v1/audio/speech. These packages are not a Chinese or Japanese dubbing tool; published voice names and latency figures are for reference only.

Gradium PhononRead story
355

HeyGen Video Goes Beyond Avatars and Generates the Whole Scene

HeyGen, best known for AI avatars, has released a general-purpose video model. HeyGen Video (API model id heygen-video-1) was announced on October 21 and the blog post was updated on October 22.

HeyGen VideoRead story
356

Railway says an attacker reached customer environment variables and some decrypted values may have been taken

Railway says an attacker accessed stored environment variables and may have decrypted some. It urges every user to rotate all keys stored on the platform.

RailwayRead story
357

Claude Security Now Keeps Scanning After the First Review and Checks Whether Fixes Land

Anthropic announced on October 21, 2026 that Claude Security now scans repositories continuously as code changes, checks findings across connected repositories, marks findings fixed when the fix ships and reopens them if the issue returns, and adds an API, webhooks and an MCP connector. It is a public beta for Claude Enterprise customers.

Claude SecurityRead story
358

ChatGPT Ads Now Reach 89 Countries, but Only on Free and Go

OpenAI expanded ChatGPT Ads to 26 more countries on October 21 (Beijing time), bringing the total to 89. Ads appear only on the Free and Go plans.

ChatGPTRead story
359

Zhipu Opens Managed Agents Beta, Free Sandbox Until October 31

Zhipu's Managed Agents (Beta) runs agents in a cloud sandbox on the GLM-5.3 models. Sandbox and Web Search/Fetch tool calls are free until October 31, 2026, while model tokens are billed pay-as-you-go and Coding Plan quota is not currently supported.

Managed AgentsRead story
360

If You Installed LiteLLM 1.107.0, Rotate Your Keys Now

Malicious LiteLLM 1.107.0 artifacts were published to PyPI, Docker Hub and GHCR on October 21. They have been removed. If you installed or ran them, treat every credential the machine could reach as exposed and pin 1.106.2.

LiteLLMRead story

From the Webmaster

Personal experiences, thoughts and notes.

View all notes