This is honestly one of the coolest parts about the AI-enabled era!
Feedback can be actioned on at lightning pace 👏
45
Been using @higgsfield and playing around with my old YouTube thumbnails to see how I could make them more interesting 👀
@FluidVoiceApp going up against the big dawgs xD
compaction has a weird design flaw 😳
the handoff summary gets written by the agent at its worst moment ~ 400k tokens deep, the most lost it'll ever be
and the fresh session trusts it blindly
retrieval here: $0.10 vs up to $1.99 compacting 👀
Compaction is a mistake.
When an agent's context fills up, tools like Codex and Claude Code compact it:
1. The model prefills the entire session again
2. It writes a summary of everything so far
3. A new session starts from that summary
I measured this on a real ~400k-token session with Opus 5.5:
→ 38 s and $1.99 when the prompt cache has expired
→ 29 s and $0.17 when it hasn't
And the summary is lossy. The exact error message, the file path, the one number you needed is often gone.
So I do something new for Mainplane
Context retrieval works the other way around. Mainplane keeps the full record of every session as a file. When the context fills up, a new session starts right away with one instruction: "You are continuing session X. Retrieve key context from it as you go."
The agent searches the old session and copies the exact lines it needs, when it needs them.
Same session, same model: 15 s and $0.10 to get back to work. It does not depend on a cache hit, and nothing is paraphrased.
This is the bitter lesson applied to agent memory. Don't hand-design what the model should remember. Give it the full record and good tools, and let it decide.
Mainplane is open source (MIT). Link in the first comment.
1
1
41
~half of forever security's customers launch coding agents from ~/Downloads 😳
a site can drop a file there with zero clicks. a planted AGENTS.md got codex to install a bad npm package 90%+ of the time in their tests
where do you launch yours from?
forever.security/blog/hijack…
38
le chonk is "open weights" but the FP8 build wants 8 blackwell GPUs, and the license makes anyone selling it as an API sign a separate deal with mistral
large 3 was plain apache 2.0 ~ open is slowly turning into "open if you're an enterprise"
1
33
terra being "4x cheaper than opus" is a short-context number ~ past 272K input tokens terra bills the whole request at 2x input / 1.5x output. opus 5.5 is flat all the way to 1M
so on big agent sessions it's $2/$7.50 vs $4/$20. more like 2-2.7x 👀
1
26
Latest video on @blocks' new Dessa voice model just went live 😊
youtube.com/watch?v=PHnqyR10…
1
1
136
What I loved from @OpenAI DevDay
- Helping people distribute their apps via their plugins and marketplace 1.2b user ecosystem
What I didn't like
- Ultrafast costing 6x usage. Cool headline, but 60tps @ 1x usage is plenty for real-world work imo.
1
49
DevDay better be good 😳
Hi,
Tomorrow we are re-opening the Pro $200 subscriptions to new subscribers, but together with it we are also changing how we calculate the usage for it. In effect, if you do the math, it will net out at half the dollar in API spend compared to the old Pro $200 plan.
Now that it's said, let me explain why this is happening and why you will still get more work done than if you were on the Pro $200 subscription one month ago.
(a) We didn't want to compromise in other ways and are committing to not reintroducing the 5h limit, so that you can fully use the weekly usage when you want.
(b) On the subscription, we guarantee that over time you always get more work done and with an increasing level of quality. This means that you will continue to get more value per dollar spent as a result of models getting more efficient and us passing down the improvements in the form of API price reductions.
(c) We don't want to put an incentive on ourselves to artificially inflate the API list prices to make it look like you are getting a lot (and workaround it through discounts, etc). Instead we want to continue to both rapidly reduce prices and increase capabilities of models on the API. This week we introduced GPT-6 Sol and GPT-6 Luna at 50% of their previous price. Over time, we see prices go low enough that it makes sense for most to buy usage as needed without there being a significant gap between what you get in a subscription and what you get in the API for a dollar spent.
(d) Tomorrow, we are adding more things to the subscription that won't draw on the usage, I won't reveal what that is yet.
I wanted to be transparent before all the big announcements tomorrow. Lots of new exciting things are coming to the subscriptions that will make it super compelling, but I wanted to make sure to share this change ahead of time so you can all understand it before we shower you with good news.
Codexingly,
Tibo
46
GPT-6 Sol is so painful to work with rn ~ just cancelled one of my GPT subs, will be moving more to Claude for Opus 5.5 😳
47
Terra fast mode is twice the price and twice as slow as normal mode? 🤔
Am I reading this correctly? @OpenAIDevs
35
Clem retweeted
Cloud AI dictation subscriptions are starting to look like a scam.
We just got accurate, formatted AI dictation running fully ON YOUR iPhone.
If we can do this for free, explain to me why you’re paying $10–20 every month.
Did @OpenAI just remove the 5hr limit? 👀
Or is this a UI bug @thsottiaux ~
5hr limit was the only thing holding me back from using GPT 5.6 more 😬
1
167











