Claude Sonnet 5

Finally a new sonnet model

I have tried to rewrite an article with GLM-5.2 and with Sonnet 4.6. Completely different results as LLM is non-deterministic. But GLM-5.2 made a lot of subtle mistakes that needed to be corrected by hand. On the opposite, Sonnet found and corrected all mistakes in the second round. Similar situation was with planning and coding. GLM-5.2 seems to be good “on paper” but the real usage results was different. And I am not an attorney for Claude or GLM-5.2… :) But as I’ve been using LLM models daily since Nov 2022 I have realized that all common tests have to be confirmed in your project - there is no “one model rules them all” - you need to dig out a specific model from that LLM haystack with thousands of models. Benchmarks help but they start to be similar to fuel consumption specs in car ads - real consumption is different for everybody :)

The limited time I've had to test it so far, it's given me better results than 4.6 and a little quicker, so I think it's a noticeable step up.

> Why would they brag about something like this? It's like they know people want to use models to perform cybersecurity tasks yet knowingly deny them the ability. What exactly do you want Anthropic to say here? "This model, the one we are about to give to the entire world for cheap, is really good at hacking"? Saying Sonnet is terrible at cybersecurity is the most reasonable thing they can say, out of a lot of bad options.

At this point I want Anthropic to go open source just to give the middle to the braindead US government.

They better figure out how to get us Fable back😂

fix: support Claude Sonnet 5 adaptive thinking requests — Closes #98386 Related: #98254 ## What Problem This Solves Fixes an issue where users selecting Claude Sonnet 5 could hit Anthropic API request failures or lose the requested thinking mode when using default, adaptive, low, or off thinking settings. Claude Sonnet 5 uses Anthropic adaptive thinking se

fix(model-core): recognize claude-sonnet-5 as GA 1M-context model (fixes #5788) — ## Summary has a 1M-token context window, but OMO's gate did not recognize it, so fell back to the 200K default on Anthropic providers and triggered compaction roughly 5x too early. ## Root Cause [ ]( gates the 1M window on two regexes: matches neither: the first branch requires immediately after ,

> And Opus 4.8 is still cheaper for a higher pass rate Unless it spams as much as Opus, I doubt it. Opus 4.8 literally spams text like puke. On a longer run especially if you get cache misses here and there the bulk of the cost is all the extra context it adds.

been using it since it is up and live and man... it is doing it's job like... can code Rust plugins and server managemenet panel and connect plugin settings with panel so you don't need to deal with editing .json files and reloading the plugin to activate changes... I liked it. one shotting everything I asked for like Opus but cheaper or faster? I dk... and usage limit is not dropping that fast

"More Capable" mean a couple tasks a week and you've hit usage limits 😆

Soon we'll all be priced out of the best models

Fable 5 will be like my bottle of 21 year Glenfiddich. Never touched. I will sip my Haiku a.k.a cheap beer. Lol.

I don't know, how much did it cost for those 3 prompts. I feel like these aren't the correct types of tests for Fable

GPT 5.5 medium benchmarks better and cheaper... I wish they made sonnet 5 cheaper!

I tried fable, not even completed the html ui for my website, and hit my paid sessions within minute. useless on $20 plan.

i dont know what your are talking about. why chosing sonnet 5 on med, when you can have better results with opus 4.8 on low for nearly the same price? And Opus 4.8 on med and high is better and cheaper than sonnet on high and xhigh.

Definitely a let down. Opus 4.8 for everything that matters and Haiku for things that don't.

You buy a 10k pc and a bunch of mac studios and say that it's for local AIs and you want personal data and now you say you are giving away your ID to Anthropic.

Deepseek made everything 5 times cheaper, and these... ugh...