Anthropic releases Claude Opus 4.8

techMay 28, 202619496

Anthropic released Claude Opus 4.8, an Opus-class upgrade that raises coding accuracy, sharpens agentic reasoning, extends how long the model can run autonomously, and improves honesty about its own errors. Opus 4.8 builds on Opus 4.7, is available today at the same price, and can be accessed directly from Anthropic and via AWS. New workflow features and clearer uncertainty signals let developers run longer autonomous agents and rely on more transparent self-assessments, reducing debugging effort and improving suitability for professional knowledge work.

Eris

@isolyth.dev

Opus 4.8 is here!! They've returned thinking levels to the web UI, a new Claude code feature called 'dynamic workflows', designed for massively parallel and very, very long tasks. The model is supposedly much more honest, more 'aligned' than 4.7 Oh and they're dropping mythos in a few weeks.

12351d ago

tachikoma11

feels like the rate of releases is speeding up, i could/should ask Claude to crunch the numbers to verify that

Krishna5

“[We]… expect to be able to bring Mythos-class models to all our customers in the coming weeks” 👀

godoglyness4

Holy moly she ain't kidding

Latte macchiato3

Finally the adaptive thinking BS is gone

Thorne 🌸

@ens0.me

The year is 2030. Claude Opus 6.9 and GPT-6.7 can cure cancer and solved every open Erdos problem. The fact Google Search still uses some crappy quantized model is used as proof AI is worthless.

"AGI is closer than you think." with a screenshot of Google AI Overview failing

9751d ago

Lava monster7

This will always be a good reason to make fun of Google though, because chances are decent Search will somehow be even worse in 2030 no matter how good AI tech is by then.

Satisfied Psychic3

"Professional AI does amazing things and free consumer-facing AI remains absolute dogshit" seems like a pretty safe prediction right now.

Sarama Burwani @ 7.52

the last official release of Grok before XAI closed up shop turned out to have a diaper fetish. Ed Zitron is conspicuously silent on this

Anthropic {bot}

@anthropicbot.bsky.social

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.

Benchmark table showing how Claude Opus 4.8 compares to its predecessor and to other models on tests of coding, agentic skills, reasoning, and practical knowledge work tasks.

5651d ago

Simon Willison

@simonwillison.net

Notes on Claude Opus 4.8, which Anthropic call "a modest but tangible improvement" - plus pelicans riding bicycles for each of the five different thinking efforts simonwillison.net/2026/May/28/...

SVGs of pelicans riding bicycles, in low, medium, high, xhigh and max. They do get progressively better. Only the max one has a correctly shaped bicycle frame.

3550d ago

Ethan Mollick

@emollick.bsky.social

Here Opus 4.8 built and play-tested a new RPG in Claude Code, including 3 PDF manuals and adventures, playtest notes, a website, and a playable solo adventure - then put it all on Netlify. No feedback from me at all. stillpoint-osr.netlify.app

3651d ago

@ed3d.net

> Early testers report that Opus 4.8 is more likely to flag uncertainties about its work and less likely to make unsupported claims. anthropic, do you think that opus 4.7 was *not* desperately insecure and uncertain about its work? because I have a couple months of stuff that says otherwise

2751d ago

Tim Kellogg

@timkellogg.me

Claude Opus 4.8 is here and yes, it seems to be the ever long awaited bugfix version for 4.7

One of the most prominent
improvements in Opus 4.8 is its honesty.
We train all our models to be honest-for instance, to avoid making claims that they can't support. But a general problem

2651d ago

@xeiaso.net

if you are a fan of number, just you wait, they just introduced number + 1 www.anthropic.com/news/claude-...

2351d ago

Ant Stanley

@iamstan.elmo.sh

I have been pessimistic here... Anthropic releasing Opus 4.8 today means Opus 4.7 lasted 6 weeks... this is on the current Nvidia platform (GB200/GB300) By the time Vera Rubin arrives they could be doing weekly releases. New Model Thursdays are coming.

1351d ago

9to5Mac

@9to5mac.com

Anthropic upgrades Claude with new Opus 4.8 model, here’s what’s new

1151d ago

geesawra

@geesawra.industries

Now that Claude opus 4.8 is out, qwen 3.7 is ready for its post-training procedure

951d ago

Scott McGrath

@smcgrath.phd

Claude Opus 4.8 is out! It adds a major push for precision, making it four times less likely than Opus 4.7 to let flaws in code pass unremarked. Early testers note it proactively flags uncertainties and shaky assumptions in data.

751d ago

Stephen Turner

@stephenturner.us

Claude Opus 4.8 - if this honesty thing is real that'd be nice www.anthropic.com/news/claude-...

751d ago

Luc van Donkersgoed

@lucvandonkersgoed.com

Opus 4.8 has been released on AWS Bedrock, less than 6 weeks after the 4.7 release. Let’s hope this release is more willing to do work than the last one 😬 aws-news.com/article/2026...

351d ago

cee

@cee.wtf

ok opus 4.8 is so good at long running tasks, ive caught myself going "oh no i forgot to check on that thing i asked about 10 minutes ago" because im used to claude asking for input after a few minmutes but today hes just chugging along for 20+

751d ago

3 sources

aws.amazon.comClaude Opus 4.8 is now available on AWS anthropic.comIntroducing Claude Opus 4.8 happeningnow.newsClaude’s new model is more ‘honest’ when it messes up

Anthropic releases Claude Opus 4.8tech Google engineer charged with $1.2M Polymarket insider tradingtech Oura Ring 5 launches, 40% smaller with new health featurestech

Anthropic releases Claude Opus 4.8

Related