Improving GPT-5.6 Sol in ChatGPT—and expanding access for free users

They must really be feeling the commoditization pressure. I'm not sure what the way out of this is, ChatGPT and Claude are still good products, but they are not necessarily premium products anymore.

I expect a few things to happen in the next year:

1) Exclusive MCP server deals/API integrations

2) Significant switch to B2B marketing, even moreso than we've seen before, with API interfaces being paid and chat-client interfaces becoming more and more free, perhaps just with limits more on integrations or data visualization/analysis

3) US restrictions on B2B contracts with non-US hosted models that do any sort of contracting with the government

Obviously there's a bunch of stuff I'm not foreseeing. But it really does feel like the bottom of the market is collapsing into free. I assume OpenAI and Anthropic think their next generation of models will restore their halo tier status and that the cash burn is justified to just get there, but this has to really mess up IPO plans.

I think it’s a misread to think the default ChatGPT model switching to GPT 5.6 Luna is some sort of desperation move. Keep in mind that Claude .ai never had this extreme stratification between the frontier models and the free tier (Sonnet is available to free users with rate limits).

So 5.6 Luna is just their next version of what they used to call 5.5 instant tier

And 5.x instant models were never much to write home about anyway so the default ChatGPT free model hasn’t been particularly distinctive since 4o

I can't wait to never see a reasoning button ever again. Why do I have to reason about what reasoning level to use?

Its always fun to try to read between the lines here to speculate why they are doing this.

Maybe Luna efficiency gain was actually significant enough that putting all the free users and giving them super generous limits makes sense.

They might be doing this to improve the messaging of AI among causal users since right now there is a huge amount of datacenter backlash in the US due to AI grievances.

Maybe they have too much excess capacity or they really want to juice token numbers and market share on their dashboards for marketing.

I also wonder if being given access to an actually a decent model like luna with actual thinking budget instead of brainless "instant" modes will start to make causal users understand the real capabilities of these models.

free unlimited luna is a pretty badass move

luna is very good

  Every week, 1 billion people turn to ChatGPT for everything from quick questions and web searches to planning, research, advice, and complex decisions.

Guess, Google's AI Mode is chipping away at their consumers (I know I haven't used Chat in a long, long while for 'quick questions and web searches' after OpenAI did away with "think" which I always use). The money-minting office & coding market Anthropic has cornered is hyper-competitive at both the frontier & low-cost ends. OpenAI is reactive [0] and seems right up against it, despite the strength of its excellent models.

[0] Won't put it past OpenAI (and/or Google) to open weight larger models!

At first I thought the Sol updates was perhaps trying to help with some complaints of Sol burning through tokens, complaints that have prompted some new data points on https://codex-resets.com/ .

But seeing the graphic with the visual weather report: that makes me think that is not the goal at all. :)

So 5.6 Luna is just their next version of what they used to call 5.5 instant tier

And 5.x instant models were never much to write home about anyway so the default ChatGPT free model hasn’t been particularly distinctive since 4o

I can't wait to never see a reasoning button ever again. Why do I have to reason about what reasoning level to use?

Some people want quick results. Some people want it to keep searching for a new math proof overnight without giving up, and they have money to burn.

It seems like giving it a time limit or a budget in dollars would be clearer, though?

Or, keep searching until I come back to the computer and ask about progress.

Because the model can't read your mind and know if you want a quick answer, or an hour long deep dive.

Because your incentives are opposed to the provider’s incentives

I think it's nice to be able to make the model reason for dozens of minutes when you want to go deep on a topic, even if the router thinks it's an easy question.

They must really be feeling the commoditization pressure. I'm not sure what the way out of this is, ChatGPT and Claude are still good products, but they are not necessarily premium products anymore.

I expect a few things to happen in the next year:

1) Exclusive MCP server deals/API integrations

3) US restrictions on B2B contracts with non-US hosted models that do any sort of contracting with the government

free unlimited luna is a pretty badass move

luna is very good

I'm not sure I agree

1) Back then, even as a free user you'd be able to use the strongest model (even if with tight limits). Now, you need to pay to use Sol, and you need to pay to use Opus or Fable. It does seem fairly premium in that sense. Idk about 5.6 Luna, but the previous Instant was really bad, even for very casual users. It would hallucinate non stop.

2) When $100 and $200 per month plans launched, they were received as outrageous even here. Nowadays they are pretty common among power users.

Any improvement to the ChatGPT free plan is really nice. It's easy to forget that most people have never used a SOTA model, and only think of their experience with GPT-4o or Google Search's AI Overviews when asked about AI.

It's also pretty cheap if you're paying for it (for coding). But I don't quite trust it for anything complicated, so I often use Terra.

I guess I should give it another shot because I had pretty bad experience with Luna when it first came out. Stuff Sonnet knew better.

Were free users subject to some limits? I've never run into any limits.

Its always fun to try to read between the lines here to speculate why they are doing this.

Maybe Luna efficiency gain was actually significant enough that putting all the free users and giving them super generous limits makes sense.

They might be doing this to improve the messaging of AI among causal users since right now there is a huge amount of datacenter backlash in the US due to AI grievances.

Maybe they have too much excess capacity or they really want to juice token numbers and market share on their dashboards for marketing.

Let’s speculate: they want luna to be their first model “on silicon” and need as much test data as possible before finalizing the design.

It's also going to be more training data for them.

Is this new behavior from them? I haven't looked at their free chat offering in a minute, but I thought they always had them close behind paid tier, often with essentially equal products that made it weird for them to sell the paid tier for chat.

> Maybe Luna efficiency gain was actually significant enough that putting all the free users and giving them super generous limits makes sense.

Definitely this. The recent 80% discount was a reaction to Deepseek's update so that they still position near the frontier. My theory: Luna has always had a much higher efficiency. You do know that the model didn't get faster after the discount?

At first I thought the Sol updates was perhaps trying to help with some complaints of Sol burning through tokens, complaints that have prompted some new data points on https://codex-resets.com/ .

But seeing the graphic with the visual weather report: that makes me think that is not the goal at all. :)

What a bizarre example of a more direct answer. Any human would simply say “No, it doesn’t rain here in summer.”

Even after identifying the 0% chance of rain, it still drags the conversation on and on and on

  Every week, 1 billion people turn to ChatGPT for everything from quick questions and web searches to planning, research, advice, and complex decisions.

[0] Won't put it past OpenAI (and/or Google) to open weight larger models!

$20/month also gives you API access for coding, in any coding agent. I keep hitting the weekly limit but it's a good deal while it lasts.

Google’s AI has been very glitchy for me lately. I used to reserve chatGPT for serious work and Gemini for daily personalized unimportant things. But now I switched completely to ChatGPT and resigned myself to their memories/personalization.

Did they do away with think? I think now you have to do it with /think

Because your incentives are opposed to the provider’s incentives

I think it's nice to be able to make the model reason for dozens of minutes when you want to go deep on a topic, even if the router thinks it's an easy question.

Some people want quick results. Some people want it to keep searching for a new math proof overnight without giving up, and they have money to burn.

It seems like giving it a time limit or a budget in dollars would be clearer, though?

Or, keep searching until I come back to the computer and ask about progress.

The _Intelligence_ part of AGI should be able to guide the user through that without all the knobs.

Because the model can't read your mind and know if you want a quick answer, or an hour long deep dive.

Then it should just ask the user what they want if its unclear from the context.

I guess I should give it another shot because I had pretty bad experience with Luna when it first came out. Stuff Sonnet knew better.

It's also pretty cheap if you're paying for it (for coding). But I don't quite trust it for anything complicated, so I often use Terra.

Terra is awful and not cheap enough to make up for it. I'd suggest using luna or sol.

I'm not sure I agree

2) When $100 and $200 per month plans launched, they were received as outrageous even here. Nowadays they are pretty common among power users.

I also scoffed at the ChatGPT $200 Pro plan back then.

Back when coding for me still meant copy-paste from the web version, it was only worth the $20/month for me.

They only added the $100 Pro plan in April during GPT 5.4 times.

Today I happily pay $400/month for Codex and Claude Code.

>1) Back then, even as a free user you'd be able to use the strongest model (even if with tight limits). Now, you need to pay to use Sol, and you need to pay to use Opus or Fable. It does seem fairly premium in that sense. Idk about 5.6 Luna, but the previous Instant was really bad, even for very casual users. It would hallucinate non stop.

This is kind of what I'm saying though. Bottom has fallen out, differentiation is just can you be much more premium than the competition. Currently that remains unanswered.

EDIT: I'm basing this off the assumption that for chat, premium is not a point of differentiation at all. For coding/analysis, it is.

Did they do away with think? I think now you have to do it with /think

$20/month also gives you API access for coding, in any coding agent. I keep hitting the weekly limit but it's a good deal while it lasts.

What a bizarre example of a more direct answer. Any human would simply say “No, it doesn’t rain here in summer.”

Even after identifying the 0% chance of rain, it still drags the conversation on and on and on

GPT models were RLHF'd to death, they will never give a final, direct answer. Every release since the GPT-4o catastrophy is like this, it's so tiresome. They need a complete reset before it can become actually usable again. Or not as maximum engagement seems to be their goal.

Would you like to know more? Seriously though, of course, it's tuned for maximum engagement, not maximum efficiency. I wonder how long this engagement dopamine circus can last... too long apparently.

No, free tier was total garbage with GPT-5.

You can't even pick 5.6 luna on the chat app with a paid subscription. It just gives you sol (or use older models) with what seems like basically as much usage as you want. And sol is considerably smarter than luna.

All of this is of pretty minor importance though. You can't read as many tokens as a subcription can produce so more chat is not the value add nor super important.

I mean there are literally so many providers for free chat if you are willing to use several seperate apps.

The real value in these subs is using codex cli, much like the real point of anthropic subs is using claude code. Because agentic work actually does require a lot of tokens.

Then it should just ask the user what they want if its unclear from the context.

The _Intelligence_ part of AGI should be able to guide the user through that without all the knobs.

Were free users subject to some limits? I've never run into any limits.

Terra is awful and not cheap enough to make up for it. I'd suggest using luna or sol.

So instead of just getting an answer, I have to wait until it asks me, and I have to type back a response? Extremely annoying.

Exactly! I posted much the same comment in another thread and there were lots of huffy complaints that amounted to "you're prompting it wrong"

That's what the reasoning slider is for!

OpenAI tried auto-routing with the initial GPT-5 release and it was immediately clear why that was a bad idea.

intelligence is not omniscience though

It used to be 10 messages every 5 hours using GPT-5, then unlimited 4o-mini. More recently, it went down to ~5 messages per day on 5.5 Instant, then unlimited on 5.5 mini.

Terra is far from awful. I've not used Luna enough to judge whether it's significantly better than Luna, but at least via GitHub Copilot, Terra is leaps ahead of Sonnet 5.

So Luna is more awful but it's OK because it's cheaper?

It's also going to be more training data for them.

Let’s speculate: they want luna to be their first model “on silicon” and need as much test data as possible before finalizing the design.

I also scoffed at the ChatGPT $200 Pro plan back then.

Back when coding for me still meant copy-paste from the web version, it was only worth the $20/month for me.

They only added the $100 Pro plan in April during GPT 5.4 times.

Today I happily pay $400/month for Codex and Claude Code.

> Maybe Luna efficiency gain was actually significant enough that putting all the free users and giving them super generous limits makes sense.

This is kind of what I'm saying though. Bottom has fallen out, differentiation is just can you be much more premium than the competition. Currently that remains unanswered.

EDIT: I'm basing this off the assumption that for chat, premium is not a point of differentiation at all. For coding/analysis, it is.

especially when 99% tasks don't need fable, and certainly not the next model

All of this is of pretty minor importance though. You can't read as many tokens as a subcription can produce so more chat is not the value add nor super important.

I mean there are literally so many providers for free chat if you are willing to use several seperate apps.

The real value in these subs is using codex cli, much like the real point of anthropic subs is using claude code. Because agentic work actually does require a lot of tokens.

No, free tier was total garbage with GPT-5.

Would you like to know more? Seriously though, of course, it's tuned for maximum engagement, not maximum efficiency. I wonder how long this engagement dopamine circus can last... too long apparently.

OpenAI tried auto-routing with the initial GPT-5 release and it was immediately clear why that was a bad idea.

intelligence is not omniscience though

That's what the reasoning slider is for!

So instead of just getting an answer, I have to wait until it asks me, and I have to type back a response? Extremely annoying.

Exactly! I posted much the same comment in another thread and there were lots of huffy complaints that amounted to "you're prompting it wrong"

It used to be 10 messages every 5 hours using GPT-5, then unlimited 4o-mini. More recently, it went down to ~5 messages per day on 5.5 Instant, then unlimited on 5.5 mini.

especially when 99% tasks don't need fable, and certainly not the next model

So Luna is more awful but it's OK because it's cheaper?

Yes. You're better off running Luna Max or Sol Medium and never touching Terra.

Terra is far from awful. I've not used Luna enough to judge whether it's significantly better than Luna, but at least via GitHub Copilot, Terra is leaps ahead of Sonnet 5.

Yes. You're better off running Luna Max or Sol Medium and never touching Terra.

Our mission is to ensure that artificial general intelligence benefits all of humanity. We’re introducing updates to ChatGPT that improve everyday conversations while expanding access for Free users.

For Plus and Pro users, we’re updating GPT‑5.6 Sol in Chat to be more reliable with facts and provide more focused answers. A new slider lets you choose how much thought ChatGPT puts into each response.

For Free users, we're updating the default model to GPT‑5.6 Luna and expanding access with unlimited text chats. For questions that need more thought, a new Think button lets you access higher reasoning for harder questions.

GPT‑5.6 Sol, tuned for how you use ChatGPT

Every week, 1 billion people turn to ChatGPT for everything from quick questions and web searches to planning, research, advice, and complex decisions. We’ve updated GPT‑5.6 Sol to better support that full range. It delivers more focused answers, adapts its level of detail to the question, avoids unnecessary formatting, and offers a helpful correction when simply agreeing wouldn’t be useful. For Plus and Pro users, the same model now powers both Instant responses and deeper reasoning, creating one consistent experience.

More reliable facts

A useful answer needs to get the facts right. The new GPT‑5.6 Sol is designed to make fewer mistakes—especially when answers depend on dates, numbers, sources, rules, or assumptions—by better using the sources it finds to answer your question.

In an internal evaluation of financial, medical, and legal prompts requiring factual detail, responses containing at least one factual error were about 62% less common with GPT‑5.6 Luna and 68% less common with GPT‑5.6 Sol than with GPT‑5.5 Instant.

More consistent, from quick answers to deeper thinking

With this update, we’re also bringing ChatGPT’s Instant and Thinking experiences closer together, creating a more consistent tone and behavior across different kinds of conversations. When you move from Instant to higher effort, it should feel like the model is taking extra time for a more comprehensive answer—not like you’re switching to a different model with its own tone or style.

Plus and Pro users can use the new slider in ChatGPT on web, mobile, and desktop to choose how much thought ChatGPT puts into an answer. Keep it quick for everyday questions, or move the slider up for planning, research, writing, coding, or decisions that need more thought.

Expanding access for free users

We’re expanding access to our latest models for free users with unlimited text chats using GPT‑5.6 Luna, plus a new Think button for harder questions.

For questions that require deeper reasoning, Free users can tap the new Think button to give GPT‑5.6 Luna more time to work through the answer.

ChatGPT Free interface showing the Think button and a “Get smarter answers” dialog with options to upgrade to Plus or turn on deeper reasoning.

Safety and availability

Plus and Pro users can access the updated version of GPT‑5.6 Sol and the new slider in ChatGPT starting today.

GPT‑5.6 Luna will become the default model for Free and Go users this week. Starting next week, they’ll also have unlimited text chats and access to a new Think button for harder questions (subject to abuse guardrails). Limits will still apply for file uploads, images and other tools.

Because this version of GPT‑5.6 Sol is optimized for everyday chats, it will only be available in the Chat experience in ChatGPT. The version of GPT‑5.6 Sol that powers Work and Codex is not changing as part of this release.

You can find more detail on safety training and evaluations in our system card⁠(opens in a new window), which outlines additional measures we’ve introduced to support users we believe are under 18. For these users, we trained the model to avoid romantic roleplay, age-restricted challenges, and presenting itself as a substitute for real-world relationships. In addition, we applied age-appropriate boundaries around sexual content, eating disorders and body-image risks, age-restricted goods, dangerous activities, and graphic violence. Finally, the model encourages connection with trusted people when a teen may need support. We reinforced this training with system-level protections and have added new evaluations for how our models perform for users under 18, and are continuing to improve model responses in this area.

More intelligence for more people

This is a concrete step toward more abundant intelligence: making our latest models more widely available, improving the usefulness and reliability of the answers people get, and letting free users keep text chats going without a rate limit. Access shapes opportunity, and this update gives more people the ability to keep asking, develop an idea, and get help when they need it.

Hacker Times