← Back to Blog

August 10, 2026 · By JayyRedd

ChatGPT Is Now Unlimited For Free | Daily AI News Brief (Aug 10)

📺 Watch the video

ChatGPT just went unlimited for free users. Europe started forcing every AI system on the continent to admit it's an AI. And Alibaba is about to give away a model it's selling at OpenAI prices. Three moves, one week, and every single one of them changes what it costs you to build with this stuff. I'm Jayy, this is your Daily AI News Brief, let's get into it.

Story 1 — OpenAI makes free ChatGPT unlimited

Let's start with the one that hits the most people. OpenAI announced that free and Go tier ChatGPT users now get unlimited text chats, running on GPT five point six Luna as the new default model. No more hitting a wall mid-conversation and getting bumped down to a weaker model. Free users are also getting a Think button, which lets you spend extra reasoning time on a single hard question instead of paying for a subscription to get it. That started rolling out this week. Now, unlimited means unlimited text. File uploads, image generation, and the heavier tools still have caps, and Sam Altman has already gone on record saying unlimited pricing may not survive long term. Here's the creator angle. If you've been paying twenty dollars a month purely to avoid rate limits on drafting and research, that reason just evaporated. Test the free tier for a week. Keep the subscription only if you actually need the tools, not the token budget.

Story 2 — Europe's AI labeling rules are now enforceable

Story two is the one almost nobody in the creator space is talking about, and it's the one most likely to land on your desk. As of August second, Article fifty of the EU AI Act is live and enforceable. That's the transparency layer. If someone is talking to a chatbot, you have to tell them it's a bot. If you publish AI generated or AI manipulated content, it has to be marked as such in a machine readable way. Deepfakes, synthetic audio, AI images, all of it. National regulators can enforce it starting now, and the penalty ceiling is fifteen million euros or three percent of global turnover, whichever is bigger. There's a grace period on the marking and detection piece for systems already on the market — that one lands December second. And content you published before the deadline doesn't have to be retroactively labeled. Creator angle: if any part of your audience is in the EU, start labeling now and build it into your workflow while it's cheap. Retrofitting disclosure across a back catalog later is a nightmare.

Story 3 — Alibaba's Qwen and the collapsing price of intelligence

Story three. Alibaba shipped Qwen three point eight Max, a two point four trillion parameter multimodal model with a one million token context window, and priced it at two dollars per million input tokens — dead even with GPT five point six. Then they announced open weights are coming this week, including a smaller twenty seven B checkpoint you can actually run locally. Think about what that means. They built a frontier class model, matched the American price, and then said we're giving the weights away anyway. That's not a product launch, that's a deliberate attempt to make high end intelligence a commodity. Creator angle: your per token costs are on a one way trip downward. Don't architect your business around today's API prices being permanent, and if you've been waiting for a local model good enough to handle your bulk drafting offline, this week is worth watching.

Story 4 — Meta ships Muse Code

Story four. Meta launched Muse Code, its first real coding agent, powered by a new model called Muse Spark one point two. It lives in your terminal, it runs sub agents in parallel on hard problems, and it keeps a log of what it did so a crash doesn't send it back to zero. It ships with commands like slash plan, which drafts an approval based execution plan before it touches anything, and slash grill, which stress tests that plan first. Pricing is pay as you go at a dollar twenty five per million input tokens. That's Meta going straight at Claude Code and Codex on price. Creator angle: if you build your own tools, funnels, or automations, the coding agent market just got a third serious competitor, and competition on price is very good news for solo operators.

Quick Hits

Quick hits. ByteDance's Seedance two point five is rolling out — native thirty second video generation in a single pass, with up to thirty reference images to lock your characters and style. It's China first for now, with the Volcano Engine API still listed as coming soon. And Anthropic confirmed it's building an in house chip design team, co designing silicon alongside Claude, targeting roughly a fifty percent cut in inference cost per token. Cheaper chips eventually means cheaper Claude.

The Takeaway

Join the Roundtable

Alright, real talk. Everything I just covered — the pricing shifts, the compliance deadlines, the new tools — that's a lot to track alone, and most of it only matters if you actually act on it. That's why I built the AI Creators Roundtable. It's a private community for creators who want to build and monetize with AI. You get a vetted library of prompts, tools, and workflows that we actually test. You get a real creator network that gives you honest feedback instead of hype. Weekly live AMAs. Exclusive job and client leads. And if you join now, founding member pricing is locked in for good.

Sources

  1. OpenAI unlimited free text chats / GPT-5.6 Luna / Think button (Aug 6, 2026)
  2. EU AI Act Article 50 transparency obligations enforceable (Aug 2, 2026)
  3. Alibaba Qwen3.8-Max + open weights week of Aug 10 (Aug 3, 2026)
  4. Meta Muse Code + Muse Spark 1.2 (Aug 5-6, 2026)
  5. Quick hits