OpenAI Drops CODEX AGENT, Manus AI New Upgrade, New Claude 3.8 Sonnet + More AI News
OpenAI Codex One: The Future of Autonomous Software Engineering Has Arrived
Alright, so a lot just dropped in the AI space, and honestly, it’s one of those weeks where everything is shifting at once. We’ve got OpenAI finally launching codecs inside ChatGPT, ManusAI stepping into image generation in a way that actually feels intelligent, Google opening up more about its search future, and Anthropic silently laying the groundwork for what looks like their biggest Cloud upgrade yet. There’s a lot going on, so let’s talk about it.
Alright, so let’s get into Codex, OpenAI’s new software engineering agent that just dropped as a research preview. If you’re using ChatGPT Pro, Team, or Enterprise, you’ve probably already seen it pop up in the sidebar. Codex works more like a full stack dev running inside a secure, cloud-based sandbox.
It operates entirely within its own isolated environment. No internet access, external APIs, nothing leaking out. You just connect it to your GitHub repo, and from there, it starts handling actual engineering tasks without needing constant supervision.
You can throw at tasks like writing new features, fixing bugs, running tests, cleaning up messy code, or even digging through your codebase to answer specific questions. It spins up a virtual environment, loads everything in, and then handles the task from end to end. That includes setting up and running tests, applying linters, checking types, the whole routine.
And the best part is, you can watch it work in real time. Permanent logs, test results, status updates, they’re all right there, so you always know what’s going on. The model running behind all this is Codex One, which is a specialized version of OpenAI’s O3, fine-tuned specifically for software development.
Codex One Raises the Bar with RL-Tuned Intelligence and a Seamless Dev Workflow
They didn’t just train it on general-purpose data, they used reinforcement learning on real coding tasks, pull request patterns, and team workflows. The result is a model that writes clean, structured code, understands project layouts, and mirrors the way human engineers actually work. In internal benchmarks, it hit 75% pass-at-one accuracy on Suibench-verified tasks.
That’s a noticeable jump over O3 High, which landed at 67%. And what’s nice is, it doesn’t need a ton of configuration to be useful. Sure, you can give it AgentsMD files to help it navigate your repo more efficiently, but even if you don’t, it still figures things out.
It respects your architecture, follows your naming patterns, and can juggle multiple tasks without stepping on its own toes. Once Codex finishes whatever you’ve assigned, it doesn’t just give you some output and call it done, it commits the changes directly within its sandbox and includes logs and references so you can trace exactly what it did and why. From there, you’ve got options.
You can review the output, make tweaks, turn it into a pull request, or pull it down locally and keep working from there. If you prefer working from the terminal, Codex CLI is probably the better fit. It’s the open-source version you run locally, and now it uses Codex Mini by default.
That’s a smaller, faster model based on O4 Mini optimized for low-latency workflows. It’s great for everyday tasks, renaming variables, writing test cases, refactoring functions, stuff that takes time but doesn’t require deep focus. You can leave it running in your terminal almost like a quiet assistant who’s always ready to help out when things get repetitive.
Codex Pricing, Ecosystem Expansion, and the Rise of Visual Intelligence with Maness AI
For API usage, pricing is pretty straightforward. Codex Mini costs $1.50 per million input tokens and $6 per million output tokens, plus there’s a 75% discount on cached prompts, so if you’re repeating similar tasks, the costs drop significantly. Codex is part of OpenAI’s push to turn chatGBT into a workspace built around task-specific agents.
There’s Operator for browsing, Sora for video, Deep Research for analysis, and now Codex for software development. Access to Codex is generous for now, though rate limits are coming. The idea is simple.
You assign real coding tasks, and Codex handles them while you stay focused. It’s built to feel like part of your team, understanding your project, following your standards, and quietly taking care of what slows you down. While OpenAI keeps expanding its agent lineup inside chatGBT, over in China, something wild just happened.
Maness AI, the autonomous agent from Monika, also known as Butterfly Effect AI, has introduced an advanced image generator that’s on a completely different level. And it’s not just another model that turns prompts into pretty pictures. It’s a full-blown visual problem solver, built right into an autonomous agent framework.
Let’s say you ask it for a modern Scandinavian living room. Maness won’t just throw together random furniture. It first analyzes your intent.
Maness AI Redefines Image Generation While Anthropic Teases the Arrival of ‘Clod 3.8’
Are you designing a catalog, creating ad visuals, or drafting a room layout? Then it builds a strategy. It uses layout engines to arrange space, style detectors to match the look, and browser tools to pull design trends or brand guidelines. It might even select real IKEA furniture, consider spatial relationships, apply color theory, and ensure everything fits the purpose.
The system is built on a multi-agent architecture, where separate modules handle planning, execution, and verification. They run independently but collaborate like a design team, allowing Maness to work through complex workflows, not just one-off prompts. That’s why it can deliver things like product campaigns, architecture mockups, or platform-ready visuals.
All consistent, brand-aware, and usable. It’s already being tested for e-commerce, product visualization, marketing content, and even architectural planning, like generating full interiors from blueprints. The big limitation? It’s still in closed beta and available by invitation only.
So unless you’re part of a select test group, you can’t use it yet. Now let’s shift to Clod. Anthropic’s been relatively quiet, but behind the scenes, they’re cooking up something huge.
There’s been a bunch of internal leaks about a new model possibly named Clod 3.8 or Clod 4. We’re seeing names like Neptune appear in their config files, and yeah, Neptune being the 8th planet is probably not a coincidence if we’re talking versioning. Publicly, they denied the rumors after that Win 4 Months of Mass contest went viral, but come on. There’s too much back-end evidence pointing to something real.
Clod’s Next Leap: Anthropic Pushes Toward True Agentic Intelligence with Transparent Autonomy
One of the standout leaks even showed internal tools with redacted model names and easter eggs tied to upcoming versions. And the information? Confirmed. Anthropic is prepping upgraded versions of both Clod Sonnet and Clod Opus.
The big thing with this new wave of Clod models is what Anthropic calls true agentic behavior. That means the model can switch autonomously between reasoning and acting without user prompts. It doesn’t just generate an answer in one shot.
It breaks down the problem, builds a plan internally, then switches into action mode to call tools, search data, or run code. If something goes wrong mid-tack, it backtracks, rethinks, and tries again. That’s a real agent, one that reasons like Gemini, but potentially with more precision in task delegation.
And this actually mirrors what OpenAI’s O3 model already does inside ChatGPT, where it can browse, run code, and iterate before showing you the final result. But Anthropic’s take might offer better transparency or control, depending on how they deploy it. For example, with the upcoming Clod update, developers might see the full breakdown of thoughts, tool calls, and revisions in the background, not just the polished final response.
And they’re not stopping there. Anthropic is also investing in making these agents work better with complex toolchains, possibly building integrations with search, databases, and APIs, all inside one flow. That’s a direct response to what Google’s doing with its own AI mode inside Search.
Google Search Enters AI Mode: Pichai’s Vision of a Conversational Future Takes Shape
CEO Sundar Pichai was recently on the All In podcast, and the big question came up whether Google is being disrupted by ChatGPT, Perplexity, and other AI-native tools that are rapidly eating into traditional search behavior. Pichai didn’t seem rattled at all. His take was that disruption isn’t inevitable unless you ignore it.
He sees it more as a shift, one that Google is actively adapting to rather than resisting. And the numbers show they’re already moving. Over 1.5 billion users have engaged with Gemini-powered AI overviews inside Google Search.
These aren’t just summaries or snippets. It’s an AI layer baked directly into search results, designed to give more context, answer follow-up questions, and reduce the need to click through multiple pages. It’s a way to keep users in Google’s ecosystem while still giving them something closer to an AI chat experience.
But they’re not stopping there. Google is preparing to launch something called AI Mode, which will turn Search into a full-on conversational experience. It’s not just query and result anymore.
You’ll be able to ask a question, get a response, follow up with more context, refine your query, and get deeper answers, all inside the Search interface. Basically, it turns Search into a Gemini-powered assistant with memory across turns. And this isn’t some distant roadmap item.
The Battle for the Smartest Agent: Google, Apple, OpenAI, and Anthropic Redraw the AI Map
It’s already been confirmed and will be showcased in more detail at Google I slash O. That said, Google’s position isn’t bulletproof. Apple recently signaled it may replace Google Search in Safari with a more AI-native system, possibly its own or powered by another provider like OpenAI. That kind of move could hit Google hard, especially on mobile, where Safari holds massive market share.
The moment that news surfaced, Google’s stock took a noticeable hit. Investors clearly understand what’s at stake here. Still, Pichai isn’t new to these shifts.
He pointed out that people had similar concerns when mobile Search took off, and again when TikTok started pulling younger audiences away from YouTube. In both cases, Google adapted, integrated features from rising platforms, and kept their core products alive. Pichai’s betting they can do the same again, this time by making Search smarter, more conversational, and more useful than anything that exists in AI chat apps today.
So while yes, Google is under pressure, especially from Apple and OpenAI, they’re not standing still. They’re building, integrating, and reshaping how Search works to stay relevant in an AI-first world. Now the question is, with Cloud, Gemini, Codex, and Manus evolving fast, who’s actually building the smartest agent right now? Let me know in the comments, drop a like if this gave you something to think about, and subscribe if you want to stay ahead of where all this is going.
Thanks for reading, and I’ll see you in the next one.
- OpenAI Drops CODEX AGENT, Manus AI New Upgrade, New Claude 3.8 Sonnet + More AI News
- OpenAI Drops CODEX AGENT, Manus AI New Upgrade, New Claude 3.8 Sonnet + More AI News
- OpenAI Drops CODEX AGENT, Manus AI New Upgrade, New Claude 3.8 Sonnet + More AI News
- OpenAI Drops CODEX AGENT, Manus AI New Upgrade, New Claude 3.8 Sonnet + More AI News
- OpenAI Drops CODEX AGENT, Manus AI New Upgrade, New Claude 3.8 Sonnet + More AI News
- OpenAI Drops CODEX AGENT, Manus AI New Upgrade, New Claude 3.8 Sonnet + More AI News
Also Read:-Â Apple Just SHOCKED Everyone: AI IS FAKE!?
