Maybe that's why Opus 4.5 has degraded so much in the recent days (https://marginlab.ai/trackers/claude-code/).
Claude Opus 4.6
101–110 of 1001 posts
Re: Claude Opus 4.6
#102Earlier quoted context omitted.
Also explains why Claude Code is a React app outputting to a Terminal. (Seriously.)
Is this a react feature or did they build something to translate react to text for display in the terminal?
They used Ink: https://github.com/vadimdemedes/ink
I've used it myself. It has some rough edges in terms of rendering performance but it's nice overall.
Re: Claude Opus 4.6
#103Re: Claude Opus 4.6
#104Earlier quoted context omitted.
Also explains why Claude Code is a React app outputting to a Terminal. (Seriously.)
Is this a react feature or did they build something to translate react to text for display in the terminal?
Not a built-in React feature. The idea been around for quite some time, I came across it initially with https://github.com/vadimdemedes/ink back in 2022 sometime.
Re: Claude Opus 4.6
#105Hmm all leaks had said this would be Claude 5. Wonder if it was a last minute demotion due to performance. Would explain the few days' delay as well.
Re: Claude Opus 4.6
#106The bicycle frame is a bit wonky but the pelican itself is great: https://gist.github.com/simonw/a6806ce41b4c721e240a4548ecdbe...
best pelican so far would you say? Or where does it rank in the pelican benchmark?
Re: Claude Opus 4.6
#1075.3 codex https://openai.com/index/introducing-gpt-5-3-codex/ crushes with a 77.3% in Terminal Bench. The shortest lived lead in less than 35 minutes. What a time to be alive!
Re: Claude Opus 4.6
#108From the press release at least it sounds more expensive than Opus 4.5 (more tokens per request and fees for going over 200k context). It also seems misleading to have charts that compare to Sonnet 4.5 and not Opus 4.5 (Edit: It's because Opus 4.5 doesn't have a 1M context window). It's also interesting they list compaction as a capability of the model. I wonder if this means they have RL trained this compaction as o…
> From the press release at least it sounds more expensive than Opus 4.5 (more tokens per request and fees for going over 200k context).
That's a feature. You could also not use the extra context, and the price would be the same.
Re: Claude Opus 4.6
#109Earlier quoted context omitted.
The cost per token served has been falling steadily over the past few years across basically all of the providers. OpenAI dropped the price they charged for o3 to 1/5th of what it was in June last year thanks to "engineers optimizing inferencing", and plenty of other providers have found cost savings too. Turns out there was a lot of low-hanging fruit in terms of inference optimization that hadn't been plucked yet. >…
> "engineers optimizing inferencing" are we sure this is not a fancy way of saying quantization?
A ton of GPU kernels are hugely inefficient. Not saying the numbers are realistic, but look at the 100s of times of gain in the Anthropic performance takehome exam that floated around on here.
And if you've worked with pytorch models a lot, having custom fused kernels can be huge. For instance, look at the kind of gains to be had when FlashAttention came out.
This isn't just quantization, it's actually just better optimization.
Even when it comes to quantization, Blackwell has far better quantization primitives and new floating point types that support row or layer-wise scaling that can quantize with far less quality reduction.
There is also a ton of work in the past year on sub-quadratic attention for new models that gets rid of a huge bottleneck, but like quantization can be a tradeoff, and a lot of progress has been made there on moving the Pareto frontier as well.
It's almost like when you're spending hundreds of billions on capex for GPUs, you can afford to hire engineers to make them perform better without just nerfing the models with more quantization.
Re: Claude Opus 4.6
#110Earlier quoted context omitted.
Also explains why Claude Code is a React app outputting to a Terminal. (Seriously.)
Same with opencode and gemini, it's disgusting Codex (by openai ironically) seems to be the fastest/most-responsive, opens instantly and is written in rust but doesn't contain that many features Claude opens in around 3-4 seconds Opencode opens in 2 seconds Gemini-cli is an abomination which opens in around 16 second for me right now, and in 8 seconds on a fresh install Codex takes 50ms for reference... -- If their m…
Why does it matter if Claude Code opens in 3-4 seconds if everything you do with it can take many seconds to minutes? Seems irrelevant to me.