What Happened
AI text generation just took a big step forward: OpenAI has released two new models, GPT-6 Sol and GPT-6 Luna, built to write and reason in a way that reads like a real person typed it, not a machine. They arrive weeks after GPT-6 Astra, OpenAI's flagship "smartest" model, and they exist for a simple reason: Astra is powerful but expensive, and most everyday writing, customer replies, and business tasks don't need — or can't afford — that level of firepower every single time.
The headline number is the price cut. OpenAI slashed API costs for both new models by 50% compared to the previous generation's pricing. Sol now runs $2 per million input tokens (down from $4) and $10 per million output tokens (down from $20). Luna, the smaller and cheaper of the two, costs just $0.10 per million input tokens and $0.50 per million output tokens.
A Quick Word on "Tokens"
If you're not a developer, "tokens" probably means nothing to you — think of a token as roughly three-quarters of a word. So a million tokens is about 750,000 words, close to the length of ten full novels. When a company says it cut the price per million tokens in half, it means generating the same amount of text now costs half as much as it did a few weeks ago.
The Numbers Behind the Claim
On independent-style benchmark tests, GPT-6 Sol running at its highest "effort" setting (essentially, how hard the model tries and how long it thinks before answering) beat Anthropic's Claude Opus 5 at Opus 5's own top setting — while costing about 9% as much per task. On a test called Agents' Last Exam, which throws long, realistic professional work tasks at AI models — the kind spanning sales, HR, finance, and support — Sol scored 56.4%, beating Opus 5's best result at 60% lower cost per task. Luna improved 5.4 percentage points over its predecessor while costing 58% less to run. On OpenAI's internal fact-checking tests, built from real conversations where users flagged the model's mistakes, Sol now makes roughly half as many factual errors as the model it replaces.
What It Means for You
This isn't just a spec-sheet update — cheaper, more accurate AI text changes what's actually worth automating. A few concrete scenarios:
At home: Drafting a firm-but-polite email to your landlord, insurer, or a customer service line used to mean staring at a blank page for twenty minutes. Now you can get a clean first draft in seconds and just tweak the tone.
At work: If your job involves writing — social captions, meeting summaries, internal memos, product descriptions — the 50% price cut means you (or your company) can run roughly twice as many drafts for the same budget, without any drop in quality.
Running a business: A small agency or shop owner can now afford to automate workflows that were too costly before: triaging support tickets, drafting sales follow-ups, or summarizing customer calls. At 9% of a premium model's cost per task, AI-assisted busywork stops being a luxury.
Studying: Students can lean on these models for study guides, essay outlines, or explaining a confusing textbook paragraph in plain language. The improved factual accuracy matters here — fewer confidently wrong answers to unlearn later.
Creative projects: Writers building a novel, game script, or short film can use the cheaper Luna model to generate dozens of dialogue or plot variations to sift through, instead of rationing every AI request.
Extra income: Freelancers offering content or copywriting services can take on more client volume at the same AI spend, which directly improves their profit margin per project.
How to Try It Right Now
You don't need a developer account or a credit card to get a feel for AI text generation. Start free, then decide if you want the paid version.
1. Start free. Head to MyKreaTool, which offers free AI text tools you can use right in your browser — no OpenAI account, no sign-up wall. Type in what you need (an email, a caption, a product blurb) and see how AI-written text actually reads before spending a cent.
2. Go straight to the source. If you want the exact models from this announcement, sign up for an OpenAI account and use ChatGPT — Sol and Luna power its faster response modes. Developers can call them directly through the OpenAI API by selecting the `gpt-6-sol` or `gpt-6-luna` model.
3. Match the model to the job. For quick, casual writing — texts, short replies, social posts — Luna is fast and nearly free. For anything with more moving parts — a business proposal, a multi-step task, research-backed writing — use Sol.
4. Compare before you commit. Run the same prompt through the free tool and, if you have API access, through Sol or Luna. Judge for yourself which reads more natural for your specific use case.
Upsides and What Changes
The practical upside is straightforward: AI-generated writing that used to be too expensive to use at scale is now affordable enough for daily, high-volume use. Businesses can automate more writing-heavy tasks without their AI bill spiraling, individuals get more usage headroom before hitting a paywall, and the text itself reads less like a template and more like something a person wrote. Response times are also faster at these lower tiers, which matters if you're using AI inside a live chat or support tool where customers are waiting on an answer.
Limitations
A few honest caveats: the benchmark numbers above come from OpenAI itself, comparing its own new models against competitors' models and settings that OpenAI chose — treat them as a strong signal, not an independently verified fact, until outside testers weigh in. "Writes like a human" doesn't mean "is always right" — these models can still state wrong information confidently, so anything factual (numbers, dates, legal or medical claims) needs a human check before you publish or send it. For your hardest, highest-stakes work, OpenAI itself still recommends the full Astra model over Sol or Luna. And if you're a casual user on a free ChatGPT plan rather than paying per token through the API, this price cut won't show up in your day-to-day experience at all — it mainly benefits developers, businesses, and heavy users paying by usage.
Conclusion
GPT-6 Sol and Luna make solid, human-sounding AI writing affordable enough for everyday use — half the price of a few weeks ago, with real gains in accuracy and task performance to match. Your one action for today: open MyKreaTool, run one piece of writing you'd normally dread — an email, a product description, a study summary — through its free AI text tool, and judge for yourself whether it reads like a person, not a bot.
Comments 0