Ed Godshaw
251 posts
Claude Opus 4.8 is out today. It's our strongest coding model yet: up on SWE-bench Pro (from 64.3 to 69.2) and noticeably more honest about its own work. It tells you when it's unsure and catches its own bugs instead of declaring victory early. Same price as 4.7.
Claude@claudeai
Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper judgment, more honesty about its own progress, and the ability to work independently for longer than its predecessors. Available today at the same price.
English
Ed Godshaw retweetledi
Nice write up from @MusicCrowns ✌️
Response has been amazing to our recent tunes, buzzing for 2026!!!
Music Crowns@MusicCrowns
‘Lost My Groove’ instantly grabs our attention with a catchy percussion arrangement that leads into a stirring soundscape and powerful vocals from The Utopiates musiccrowns.org/new-music/lost… @TheUtopiates
English
Introducing our latest update to Gemini 2.5 Pro (06-05), which we expect to become our long term stable release. At a glance:
- SOTA on HLE, Aider, and GPQA
- Now supports thinking budgets
- Same cost, on pareto frontier
- Closes gap on 03-25 regressions
English
Should children at nursery have screen time? bbc.in/4khZD0S
English
@OfficialLoganK @TheAhmadOsman @jshearin01 @OfficialLoganK for image based OCR 05-06 is completely broken. It burns thinking tokens until max completion tokens are exhausted, and then errors every time. Developer goodwill from Gemini 2.5 is fast eroding.
English
@TheAhmadOsman @jshearin01 Great feedback, would love to chat more about what your use case is (sending a DM).
In general, we are trying to make these changes minimally disruptive. But at the scale of Gemini 2.5 Pro, we cannot keep too many versions around, lots and lots of TPU's required.
English
The new Gemini 2.5 Pro update made it so much worse than the March version, just like how the January version of Gemini Flash Thinking was so much worse than the December version
Why does Google keep doing this? :(
I want the March version back please, @OfficialLoganK
Ahmad@TheAhmadOsman
Is it just me, or has Gemini Flash Thinking gotten worse recently?
English
@OfficialLoganK @OfficialLoganK this model is significantly worse than 03-25 for OCR, and as of this morning the upgrade is being forced on us despite severe bugs and regressions. What is the point of versioned checkpoints if you force upgrades on us with no warning?
English
Gemini 2.5 Pro just got an upgrade & is now even better at coding, with significant gains in front-end web dev, editing, and transformation.
We also fixed a bunch of function calling issues that folks have been reporting, it should now be much more reliable. More details in 🧵
English
❤️ WE NEED A GREAT SUPPORT BAND! ❤️
We’re looking for your suggestions for your favourite bands that might be up for supporting us in massive venues for our massive tour later this year!
Tag your favourites!
English
Ed Godshaw retweetledi
Yo! The Utopiates ( @TheUtopiates ) released their banging debut album today. Many of you will recognise them as an awesome support bands from our last tour! Fans of Kasabian & Primal Scream will love this. Give it a spin this weekend for King & Country x
open.spotify.com/album/6qUMJDja…
English
Ed Godshaw retweetledi
Fantastic set from one of the hottest up and coming bands @TheUtopiates last night at @RoughTrade. Really nice chatting you all afterwards, hope you all enjoy this journey, now let's get 'ups and downs' played on @steve_lamacq @BBC6Music. The soundtrack to summer #TheSunAlsoRises
English
@StuJLans @OpenAI @rikudotai seeing this too, seems to be a bug on their side as our workloads are well within the usage limits for the paid tier
English
Are @OpenAI throttling API accounts for GPT-3 now? Noticing numerous 429 errors on user accounts for @rikudotai. Would be nice to get some consistency on this. Businesses rely on API and 429 errors weren't a thing until recently.
English
@fedjabosnic @OpenAI yeah seeing this too across multiple accounts we're testing with
English
We're seeing lots of 429 errors (throttling) from @OpenAI even though we're only sending a couple of small requests (max 3 per second at 100 tokens each).
Anyone else seeing this?
English
Ed Godshaw retweetledi
@theclauseuk @Spotify @theclauseuk @TheUtopiates @bigimage wanted to listen to new music and so I did! Big night on the 23rd to Celebrate
English
Ed Godshaw retweetledi
Ok.. how about some really, really, really, really overdue Saturday Tunes on a Sunday? In 10 ish..
English
Ed Godshaw retweetledi
Retweet if you think it's time for a Labour government.
English
Ed Godshaw retweetledi
Unbelievably, our officers have just had to smash the window of a vehicle to get a dog out at the RAF museum Hendon. 31.5 degrees! JUST DON”T TAKE DOGS OUT IN THIS HEAT.
English
Ed Godshaw retweetledi
Deal's in the bag. Time to finish the album.
'My childhood was small.....' 🎶
Walthamstow, London 🇬🇧 English
Ed Godshaw retweetledi
Give it the fuck up Nadine. Your Starmer obsession has gone beyond anything remotely reasonable or even sane. Go and do some work. Find out about Channel 4 properly or sort out visas for our touring artists or reply to your constituents’ letters. You’re being mad and undignified.
English
Ed Godshaw retweetledi
I see @BBCNews are having a news blackout on Tory peer Michelle Mone having her house raided by police. Please retweet if you consider this to be newsworthy.
English