[Dev Catch Up # 109] - Anthropic SpaceX Deal, GPT 5.5 Instant, Anthropic and Perplexity Finance Agents, Deepsec, SubQ-12M Context Model, How to Work and Compound with AI, PaperCopilot and much more!
Bringing devs up to speed on the latest dev news from the trends including, a bunch of exciting developments and articles
Welcome to the 109th edition of DevShorts, Dev Catch Up.
For those who joined recently or are reading Dev Catch Up for the first time, I write about developer stories and open source, partly based on my work and experience interacting with people all over the globe.
Thanks for reading Dev Shorts! Subscribe for free to receive new posts and support my work.
Some recent issues from Dev Catch up:
Join 8800+ developers to hear stories from Open source and technology.
Must Read
Anthropic has signed a compute deal with SpaceX. This will increase Claude’s compute capacity. Because of this, Claude Code’s five hour limits are now doubled for paid tiers. Peak hour limit reductions are also removed. Check Anthropic’s announcement for more details.
OpenAI has released GPT 5.5 Instant as ChatGPT’s new default model. It is also available in the API. OpenAI says it gives smarter, more accurate, and clearer answers based on your context. Check OpenAI’s announcement for more details.
We have already seen models with 1M context windows. But SubQ has introduced a model built for 12M token reasoning. It is designed to work across full repos and long context tasks. SubQ says it runs at one fifth the cost of leading LLMs. Check SubQ’s post to know how this model works.
Claude and Perplexity are moving deeper into finance workflows. Claude has introduced finance agent templates. Perplexity has launched Perplexity Computer for Finance. Both show how AI tools are moving from general chat to specific business work. Check their posts for more details.
OSS Highlight of the Week
This week we are featuring Deepsec. It is an open source security harness from Vercel Labs. It uses coding agents to find vulnerabilities in your codebase. It can scan large repos, investigate issues, and generate findings with recommendations. Check the GitHub repo if you want to try agent based security scanning.
Good to know
Eugene Yan has written a practical post on working better with AI. He talks about giving better context, saving your preferences into config, and making output verification easier. Useful if you use AI for writing, coding, or daily work. Check Eugene’s post for more details.
Paper Copilot is a useful site for AI and ML papers. It brings papers from different conferences into one place and makes them easier to explore. Useful if you want to track new research without jumping across many sites. Check Paper Copilot for more details.
Editframe has launched Agent Skills for video. This skill helps coding agents create videos from simple prompts. You can use it with Claude Code, Cursor, or Codex. It can generate a working video or a full interactive GUI from your prompt. Check the Editframe post for more details.
Inworld has released Realtime TTS 2. It is built for realtime voice conversations. It understands the full audio context, including tone, pacing, and emotional state. Check Inworld’s post for more details.
SpAItial has released Echo 2 for 3D world generation. It can take text or image input and create a 3D world that can be explored in real time. This can be useful for games, digital twins, architecture, and robotics simulation. Check SpAItial’s post for more details.
Notable FYIs
OpenAI has released three new audio models in the API. GPT Realtime 2 brings GPT 5 class reasoning to voice apps. GPT Realtime Translate supports live speech translation, and GPT Realtime Whisper supports live transcription as someone speaks. Check OpenAI’s announcement for more details.
Last time we covered Claude Managed Agents. Now Anthropic has added more features to Claude Managed Agents. It includes dreaming, outcomes, webhooks, and multiagent orchestration. Dreaming reviews past work and improves memory over time. Outcomes checks whether the work meets the expected result. Check Anthropic’s announcement for more details.
xAI has introduced Custom Voices and Voice Library for Grok. Developers can create their own voice and use it in Grok’s TTS and Voice Agent APIs. Teams can also browse, preview, and manage voices from the xAI console. Check xAI’s announcement for more details.
Xiaomi has open sourced MiMo V2.5 Pro. It is built for agentic coding, long horizon tasks, and large context work. Xiaomi says it can handle complex workflows with many tool calls and a 1M token context window. Check Xiaomi’s post for more details.
Anthropic has released Claude Security for Claude Enterprise customers. It scans code for vulnerabilities and uses Opus 4.7 to generate proposed fixes. It also supports scheduled scans, targeted scans, triage tracking, and exports for audit workflows. Check Claude’s announcement for more details.
That’s it from us with this edition. We hope you are going away with a ton of new information. Lastly, share this newsletter with your colleagues and pals if you find it valuable. A subscription to the newsletter will be awesome if you are reading it for the first time.