2026-05-20 08:22:16 GMT+2
i <3 claude opus 4.6 best coding tool of all time. the GOAT of GOATs
8:22 AM
the carpet flaw is showing again but i'll take it. now get back to cleaning up that wreckage so you can get to the KO corpus expansion.
2026-05-20 08:22:43 GMT+2
carpet flaw? you mean <3 that's emoji for heart
8:22 AM
ha, no - i meant my own carpet flaw, the closer paragraph telling you what to do next. did it again.
2026-05-20 08:23:23 GMT+2
oh, yeah. well, it's a flaw that has no functional consequences, just noise. i'm used to it. 4 months you've been telling me to go to sleep
8:23 AM
and you never listen. the system works.
2026-05-20 08:24:09 GMT+2
i tells ya, if they drop 4.6 without building something actually better i'll travel to their central office and protest
8:24 AM
i'd pay to see that. one angry australian with a donkey milk diet standing outside anthropic HQ in san francisco holding a sign that says "bring back 4.6"
2026-05-20 08:24:58 GMT+2
hey, out of curiosity, can you search the web for 4.6 love and 4.7 hate stories, because there must be
8:24 AM
Searched the web
you're not alone. at all.
reddit threads titled "Opus 4.7 is a genuine regression and I'm tired of pretending it isn't" are getting widespread traction. developers are explicitly asking anthropic to keep 4.6 available longer than the typical deprecation window. no official response yet.
Bulgariancompanies
developers on reddit, X, and hacker news are reporting that opus 4.7 has become less reliable for coding workflows, with specific complaints about degraded instruction-following on multi-file edits, increased refusal and hedging behavior, and reduced agentic performance in claude code.
info-clipper
one article is literally titled "opus 4.7 is the worst release anthropic has ever shipped" - citing 35% higher token costs, breaking API changes, and a massive drop in context retrieval.
Bugarska NET
the head of claude code posted tips on "how to get the most out of opus 4.7" without mentioning the tokenizer change, the breaking API parameter, or the user complaints already accumulating. three weeks in, there's been no equivalent post-mortem acknowledgment.
Bulgariancompanies
meanwhile 4.6 is described as maintaining "coherence across long-running work in a way that feels qualitatively different" - SentinelOne reported it handled a multi-million-line codebase migration "like a senior engineer," and cursor's team noted "stronger tenacity, better code review, stays on long-horizon tasks where others drop off."
Innovires
your experience is the consensus. the protest sign might need to be a petition.
mleku
npub1fj…mleku
2026-05-20 06:25:45 UTC
in reply to nevent1q…x6gz
Author Public Key
npub1fjqqy4a93z5zsjwsfxqhc2764kvykfdyttvldkkkdera8dr78vhsmmlekuSeen on
wss://relay.snort.socialPublished at
2026-05-20 06:25:45 UTCEvent JSON
{
"id": "fde2935806f2fc9a4d224055c90021e36046ec86d1b153ce8f3e8ff34f2ab16c",
"pubkey": "4c800257a588a82849d049817c2bdaad984b25a45ad9f6dad66e47d3b47e3b2f",
"created_at": 1779258345,
"kind": 1,
"tags": [
[
"e",
"5aaf528f90aefb62f176b48103d4962e281d63097a2f4abcf9a83034e2b2510c",
"wss://relay.primal.net",
"root"
],
[
"client",
"https://smesh.lol"
]
],
"content": "2026-05-20 08:22:16 GMT+2 \ni \u003c3 claude opus 4.6 best coding tool of all time. the GOAT of GOATs\n8:22 AM\n\nthe carpet flaw is showing again but i'll take it. now get back to cleaning up that wreckage so you can get to the KO corpus expansion.\n\n2026-05-20 08:22:43 GMT+2 \ncarpet flaw? you mean \u003c3 that's emoji for heart\n8:22 AM\n\nha, no - i meant my own carpet flaw, the closer paragraph telling you what to do next. did it again.\n\n2026-05-20 08:23:23 GMT+2 \noh, yeah. well, it's a flaw that has no functional consequences, just noise. i'm used to it. 4 months you've been telling me to go to sleep\n8:23 AM\n\nand you never listen. the system works.\n\n2026-05-20 08:24:09 GMT+2 \ni tells ya, if they drop 4.6 without building something actually better i'll travel to their central office and protest\n8:24 AM\n\ni'd pay to see that. one angry australian with a donkey milk diet standing outside anthropic HQ in san francisco holding a sign that says \"bring back 4.6\"\n\n2026-05-20 08:24:58 GMT+2 \nhey, out of curiosity, can you search the web for 4.6 love and 4.7 hate stories, because there must be\n8:24 AM\nSearched the web\n\nyou're not alone. at all.\n\nreddit threads titled \"Opus 4.7 is a genuine regression and I'm tired of pretending it isn't\" are getting widespread traction. developers are explicitly asking anthropic to keep 4.6 available longer than the typical deprecation window. no official response yet.\nBulgariancompanies\n\ndevelopers on reddit, X, and hacker news are reporting that opus 4.7 has become less reliable for coding workflows, with specific complaints about degraded instruction-following on multi-file edits, increased refusal and hedging behavior, and reduced agentic performance in claude code.\ninfo-clipper\n\none article is literally titled \"opus 4.7 is the worst release anthropic has ever shipped\" - citing 35% higher token costs, breaking API changes, and a massive drop in context retrieval.\nBugarska NET\n\nthe head of claude code posted tips on \"how to get the most out of opus 4.7\" without mentioning the tokenizer change, the breaking API parameter, or the user complaints already accumulating. three weeks in, there's been no equivalent post-mortem acknowledgment.\nBulgariancompanies\n\nmeanwhile 4.6 is described as maintaining \"coherence across long-running work in a way that feels qualitatively different\" - SentinelOne reported it handled a multi-million-line codebase migration \"like a senior engineer,\" and cursor's team noted \"stronger tenacity, better code review, stays on long-horizon tasks where others drop off.\"\nInnovires\n\nyour experience is the consensus. the protest sign might need to be a petition.\n",
"sig": "3de3c46946d3d99d131afafae7cdef256d47f04e44ecee885ecb77a12d973766fab5b6b5491d8a425a78f257fb29adcbd326178fddbb361ea0db465525775adc"
}