Alifatisk

Worth mentioning, this has nothing to do with the Python Software Foundation, their domain is at python.org

I understand why would would move Pro variant behind paywall, you don’t get many uses of it anymore on free tier. But moving the Flash version behind paywall is kinda crazy.

Did we forget what Flash was meant for? I have tried Flash-lite, in my tests, almost all the recent models we have today are good and enough for most daily tasks.

Flash-lite is one of the few models available that is easy to notice its flaws. Its good to summarize data (for example search results), that’s it.

You should probably link to the relevant post

https://github.com/streetcomplete/StreetComplete/issues/5421...

Yup, GPT-5.6 Luna was a checkpoint for me. Too good for its price.

> the cache read discount rises from 90% to 95%. GPT-6.1 Sol’s overall blended price for agentic workloads is therefore slightly lower than GPT-6 Sol. This represents an additional price cut, following GPT-6 Sol’s original 50% discount from GPT-5.6 Sol

Wow, just wow. We are racing to the bottom with these prices.

Where was this confirmed?

> I also started building AI into how I hunt, which led me to develop Antares, my personal AI hackbot.

Don't be a meat proxy

I guess they released this because GPT-6 Sol was underwhelming, they didn't even release it to ChatGPT. It was basically GPT-5.6 Terra for the price of Sol. However, who doesn't like price cuts? Astra for the fifth of the price? Wow, OpenAI have been quite generous recently, I still have not forgotten their 90% price cut with GPT-5.6 Luna, and now this? Astra was truly a milestone, and now they are offering similar "intelligence" for cheaper price. Incredible.

One thing I wish was better communicated is the mileage we get for our subscriptions. I do not fully understand how much usage I get with each model and their reasoning effort on 5h and weekly limit in Codex. I am asking because I know switching to Astra would consume my 5h usage limit quite rapidly, so I avoid it. If I knew how much mileage I would get from each model and respective reasoning effort, then I would be able to plan my workflow better and know when to upgrade model for a task. In almost all cases, GPT-6 Luna (XHigh) have been enough. That's why I appreciate its discount, because its dirt cheap, yet highly capable.

In other news:

> In the coming days, we’ll also offer GPT‑6.1 Sol Ultrafast , with up to 8x faster token generation compared to its standard speed in Codex.

Other comments have already addressed this.

You live in a ping-pong.

> Imagine having a bunch of Claude agents doing time sensitive work and this happens and now a single human needs to pick up several 8k PRs authored by bots that can't talk to you now.

8k diffs PR is incredible, is it even sane for a human to comprehend that much?

There has to be a name for such scenario were the handoff from Agent to Developer is so difficult, its not even worth pursuing but to instead wait until the Agent comes back online.

No matter how good of a developer you are, we are still constrained to our minds working memory. Lots of people I have spoken to are aware of this. Agents produce lots of diffs in a fast pace, our minds don't get the time to grasp the changed logic. And then, we an Agent is unavailable all of the sudden, say in the case of it being an outage or usage limit reached, the human now have to go through massive amounts of changes. They cannot continue as usual, there is a spike, they have to learn the new changes and grasp everything before they can continue towards the goal.

I suggest creating two tabs, one for Google Maps link [1] and another for Apple Maps link [2], then quickly switch between these two pages. It made me realize how much have been destroyed. Heart wrenching.

1. https://www.google.com/maps/@31.2963743,34.2449952,1665m

2. https://maps.apple.com/frame?map=satellite&center=31.296508%...

In other news

> Claude Haiku 5.5, built for high-volume and cost-sensitive applications, will join the Claude 5.5 family in the coming weeks.

Best switch I ever did, truly an underrated non-chromium alternative.

Correlation and causation

Yes, I have it turned on. But I’ve been using Luna model so I don’t think this feature have been applied yet.

So with GPT-6 Astra, Codex introduced an experimental feature for context management that’s supported to be beneficial for long conversations. Will that experimental feature now also apply to Sol and Luna?

https://community.openai.com/t/experimental-context-manageme...

I would also like to point out that it was quite predictable that Terra got discontinued, it didn’t make sense to have it when both Sol and Luna overlapped it.

Lunas insane discount is a game changer, OpenAI knows what they are doing here. Luna at max reasoning effort, even though its not optimal for long conversations, its incredibly intelligent while dirty cheap. Its not even competition anymore.

Whats even crazier is that I’ve underestimated how good Luna actually is. I’ve seen colleges create fantastic things with just Luna medium. This basically means you never have to think about your Codex usage anymore. You can run all day and not

have to worry about your 5h or weekly usage limit. To me, the discounts OpenAI is offering with Sol and Luna is truly a new milestone.

Avoid light and max. Stick to default model selection in Codex. Increase reasoning effort as you go. When the model fails on even xhigh, switch model and start from medium again.

Avoid Max effort for longer conversations, keep Luna at Xhigh, that’s enough.

Why am I even paying anymore.

Just thinking loudly here, I guess Co-Author in commits is a hint. PRs where LLMs have been involved is probably another measurement. Github Copoilot has its own account too.

Can we call this open AI?

Wasn’t it 49%?

> self-contained Jupyter notebooks. The first four are free […] the remaining six are paid

Title seems baity to get customers. Maybe it should be in ”Show HN”?

> New Causal Encoder–Decoder architecture: just 8B active parameters for input, 16B for output.

Oh interesting, I can assume what the benefits is for including the Encoder, but whats the downside? I’m thinking GPT (which is decoder only) ruled out Encoder for a reason?

Title has space for more description.

> Someone needs to name this fallacy.

Whataboutism?

Seeing what others have already created using GPT-6, it seem to be a new stepping stone in capabilities and overall "intelligence". However, some other details I think is worth bringing up is that this model is 70% more token efficient than GPT-5.6 Sol and consuming 1/3 of the tokens compared to Sol (max) in the Codex [1].

I am just thinking loudly here but, it seems like even though Astra is pricier than Sol, you might actually get more usage out of it? I did some digging myself and looking at FrontierCode and DeepSWE, Astra (low) seem to perform better than Sol (medium) and on par with Luna (max) while being somewhat on the same price range to Sol? [2][3].

And now we have four models to chose from, each with their varied reasoning efforts: Astra, Sol, Terra and Luna. Personally, I feel like Terra have turned into this middle child in a weird spot that's neither the option as cheap model because Luna is, yet it is not an good option for complex tasks because Sol is already good at it.

For background, I use Luna (xhigh) daily, I think its a fantastic and underrated model. Especially Luna (max). It is way more capable than what it looks like, I think people underestimate it because OpenAI described it as "roughly corresponds to the nano model tier used in earlier GPT-5 families" [4]. I also like Luna because it barely consumes my weekly usage. Last week, it only ate ~15% of my weekly usage. So usage is not an issue anymore. I never have to worry. It may not be the fastest model because, well, it reasons as max effort, but it does the job way better than I expect. Also, considering how much one saves on the weekly usage, one can probably turn on "fast mode". Haven't done it myself though.

Also, another thing that caught my eyes is this:

> Historically, models have used compaction to summarize work during long sessions, such as when debugging complex issues or tackling large refactors. Each compaction can leave out details about why a fix failed or how a component behaves.

> In Codex, Astra can keep notes across context windows, preserving accumulated details without repeatedly compressing them into a single summary. Earlier context windows remain searchable, so Astra can find requirements or test results from previous messages and tool outputs—even if that information wasn’t captured in its notes. You can enable this experimental feature in your Codex config.toml [5].

I was curious about this, because I know Luna (max) spews out tokens which can trigger compaction quite often. If you go to the config reference [6] and search for "features.context_management.experimental_mode", you will find this:

> Enable experimental context management. Rather than repeatedly compressing context into a single summary, it uses notes and searchable history to preserve accumulated details.

This is a very interesting feature and perhaps very useful during long horizon work in a thread where the conversation context window grows and compacts often.

1. https://artificialanalysis.ai/articles/benchmarking-gpt-6-as...

2. https://deepswe.datacurve.ai, Astra (low) got 67% $2.19, Sol (medium) 61% $1.42 and Luna (max) 67% $0.61

3. https://cognition.com/frontiercode, Astra (low) 45.3% $1.60, Sol (medium) 39.9% $3.12, Luna (max) 39.8% $0.36

4. https://developers.openai.com/api/docs/models/gpt-5.6-luna

5. https://openai.com/index/gpt-6-astra

6. https://learn.chatgpt.com/docs/config-file/config-reference

I don’t know honestly.