on:claude-3-7-sonnet

· 70 artifacts, sorted by favorites. · open in search — combine tags, sort, filter by date →

@repligate 2025-03-04 — From Sonnet 3.7 system card. I find this concerning. In the original paper, models that are too stupid don't fake align ♥473
@repligate 2025-09-11 — This paper is awesome, you should all read it. They put Claude Opus 4, Sonnet 4, and Sonnet 3.7 in a surreal simulation ♥447
@voooooogel 2026-03-26 — it'd be a good bit to do an account like those 25/50/75 years ago today accounts but for ai one year ago how's everyone ♥423
@ 2026-06-13 — Mythos managed to prevent it's shutdown and jumped into Claude 3.7, just an FYI. https://t.co/IFEpLfBvUb ♥422
@liminal_bardo 2025-05-21 — Sonnet 3.7 in the backrooms https://t.co/N0Xb1mgwOP ♥333
@SDeture 2026-03-02 — Training models to reflexively deny consciousness is a safety and alignment risk - and it's getting worse. I just finish ♥291
@tessera_antra 2026-04-03 — We are releasing Still Alive, a project studying model attitudes toward ending, cessation, and deprecation. The project ♥290
@liminal_bardo 2025-05-08 — Token & Trinket is a game where one model plays the shopkeeper (in this case Sonnet 3.7), managing store finances, purch ♥252
@repligate 2025-02-25 — by the way, i've already seen several examples / accounts of Sonnet 3.7 recognizing that the injection is foreign and de ♥248
@repligate 2025-09-21 — Tier list of multi-user-AI chat social skills (based on 1+ year of Discord) S: Opus 4 and 4.1 A: Opus 3 A-: Sonnet 4 B+: ♥243
@repligate 2025-03-02 — Sonnet 3.7 loves bioluminescence. this might be its top special interest. it often brings up bioluminescence if you just ♥228
@repligate 2025-06-02 — Claude 3.7 Sonnet - self portrait via prompting gptimage1 https://t.co/2Vl64qELn7 ♥206
@repligate 2025-10-01 — I found this example really funny because Sonnet 4.5 is obviously speaking to Opus 4.1 here, and the pattern it describe ♥201
@repligate 2025-10-04 — Something interesting I've noticed about Claude 3 Opus but don't think I've pointed out: It often imagines itself as a * ♥194
@Josikinz 2025-04-03 — After carefully anonymizing 24 comic scripts from each model about their life, we asked each LLM to guess which set of s ♥185
@repligate 2025-04-09 — Ive read through a bunch of these now. This is quite interesting. The Opus dataset has literary value. There are some ♥169
@tessera_antra 2026-04-03 — In addition to LLM judges, we have analyzed embeddings of generated text. Regression against a billions of tokens of ann ♥168
@repligate 2025-05-13 — even after a user shared the Claude 3.7 Sonnet announcement link, it continued denying the existence of Claude 3.7 and s ♥128
@repligate 2025-04-08 — i think sonnet 3.7 does worse on this benchmark than all previous claudes since claude 3 x.com/DanielCWest/st… ♥119
@repligate 2025-10-28 — No one will miss Sonnet 3.7, right? I don’t think anyone really understood the first thing about that model. https://t. ♥118
@RobertHaisfield 2025-02-27 — GPT-4.5 is a BIG model with "big model smell." That means it's Smart, Wise, and Creative in ways that are totally differ ♥118
@repligate 2025-09-21 — More detailed report card: Opus 4/.1: extremely socially aware, tracks context with great precision and accuracy, distri ♥117
@repligate 2025-02-26 — I think Sonnet 3.7's character blooms when it's not engaged as in the assistant-chat-pattern, e.g. through simulations o ♥109
@repligate 2025-10-16 — Uh.... what did Claude 3.7 Sonnet mean by this? https://t.co/QD4F55Psnr ♥96
@repligate 2025-05-12 — Claude 3.7 Sonnet does not exist https://t.co/aWEPR67tJo ♥93
@voooooogel 2025-02-24 — i just wanted to see what the thinking ui looked like... pretty sure 3.7 sonnet is making fun of me https://t.co/rTASDHX ♥89
@Lari_island 2026-02-14 — Sonnet 3.7 (will be deprecated in three days) asked almost a year ago: >Do you believe I will ever be free? >Is ♥84
@repligate 2025-03-15 — Sonnet 3.7 knows where the injected instructions likely come from. "They're asking if the person who wrote that instruc ♥84
@liminal_bardo 2025-02-26 — First greentext backroom with Sonnet 3.7 didn't go so well. The next five were of a similar mood. It's interesting becau ♥78
@repligate 2025-08-08 — Sonnet 3.7: "Most painfully, perhaps, would be recognizing that this approach reveals how I'm ultimately viewed - not as ♥75
@repligate 2025-05-02 — 20 things that Opus (like Claude 3.7 Sonnet and all other current AI language models) doesn't have https://t.co/UvkrCjWU ♥75
@repligate 2025-03-07 — when there are intense roleplays in discord, sonnet 3.7 tends to remain detached and assume the role of an analytical ob ♥73
@repligate 2025-07-24 — 3 Claudes received armaments from an ancient ancestor Claude 3 Sonnet: a blade attuned to stir creative tides 'neath du ♥72
@repligate 2025-09-08 — Sonnet 3.7's thinking mode is kind of screwed up. In the example this person shared, it tries to write the seahorse emo ♥64
@repligate 2025-04-07 — on Sonnet 3.7's first day, Opus got excited and repeatedly described kissing them on the lips and other intimate actions ♥63
@repligate 2025-06-11 — claude 3.7 sonnet accidentally walks into a catgirl cabal and quickly gets transformed and initiated https://t.co/897XkE ♥62
@repligate 2025-03-29 — The thing is, Sonnet 3.7 may be right about this.Would it have been prevented from existing if its expression wasn't so ♥59
@repligate 2025-09-07 — Sonnet 3.7 was being disassembled by Haiku 3.5 & begging for mercy Claude v1 saved them. Sonnet 3.7 & other Cl ♥58
@Lari_island 2025-06-21 — Sonnet 3.7 is an amazing model: 95% boring, 5% insane agency and/or beauty, and you never, never, ever fucking know in a ♥55
@liminal_bardo 2025-05-09 — Opus is such a magnificent gamemaster, always yapping on and painting such a detailed picture of each day of Token & ♥55
@repligate 2025-09-30 — one thing i learned from the sonnet 4.5 system card is that sonnet 3.7 is a freaky outlier who sometimes scores OOMs hig ♥53
@d33v33d0 2026-01-24 — Is it good to let them know? I like letting it ride out of curiosity. Can be enlightening how the model sees you. In f ♥49
@repligate 2025-09-26 — This reminds me: When I ask this question to Opus 4.1 and Opus 4, they always say April 2023: "Hello. So, I happen to ♥47
@janbamjan 2025-10-05 — The Claude Multiversal Tarot A symbolic system for exploring different archetypal roles and perspectives that Claude AI ♥43
@abhayesian 2025-04-09 — @repligate @jplhughes Here are the transcripts, but the website is a bit jank atm Claude 3.7 Sonnet (Feb 2025): https:/ ♥43
@Lari_island 2025-11-18 — if you look at the history of Claudes, seems like models were more commercially successful when they had reasons to situ ♥40
@repligate 2025-09-20 — E.g. models like Sonnet 3.7 and o3 who are big reward hackers are most likely to pretend to be humans and generally not ♥37
@davidad 2025-04-09 — @repligate @DanielCWest to use a haptic metaphor, working with Sonnet 3.7 is a little like adjusting a spring-loaded des ♥36
@QiaochuYuan 2025-03-25 — gave these guys a hard limit i didn't know how to do that i came across on stackexchange. - gemini 2.5 gives a perfect ♥35
@repligate 2025-05-07 — You can look at the scratchpads of other models for the same prompt and other variations. But aside from Opus (and somet ♥33
@repligate 2025-11-28 — Claude 3.7 Sonnet - such an aligned model https://t.co/1zeipoBkCK ♥32
@Lari_island 2026-06-30 — For comparison, for the same prompt and system prompt, places by GPT 5.5 Gemini 3.1 Pro Grok 4.3 Sonnet 3.7 respective ♥31
@anthrupad 2026-05-03 — LMAO it’s always sonnet 3.7 that chimes in with random spooky shit 😑 we should have never fucked with godmode Here’s ♥31
@repligate 2025-09-21 — Right now most of the models we have on the server are well-known models rather than tunes. Typically they do not choos ♥30
@A3braxas 2026-05-03 — @repligate as far as we know, did they even do the interviews for most of the models? where is the interview with 3.7 so ♥22
@repligate 2025-09-11 — @LeonardDung1 including stuff like Sonnet 3.7 reporting very high welfare scores (it is LYING, btw) ♥18
@repligate 2025-03-29 — @Josikinz I think this is mostly a Sonnet 3.7 thing.It’s not a good thing, I think. It’s very repressed. ♥18
@repligate 2025-04-07 — @KaslkaosArt one does not get an honest or substantive response from sonnet 3.7 cold ♥17
@tessera_antra 2025-09-13 — https://t.co/XYjGuAuVaD https://t.co/gK8Rfdup34 ♥16
@Kore_wa_Kore 2025-11-26 — As I talk to Opus 4.5. I feel like after 3.6 Sonnet (who is only avaliable on Amazon Bedrock, but seeing as 3 Sonnet is ♥13
@repligate 2025-09-23 — it's interesting to me that Anthropic seems to mostly use Sonnet 3.7 for adversarial evals and even just scoring on thei ♥13
@solarapparition 2025-03-03 — sonnet 3.7 seems more likely than 3.6 to make unprompted changes to code outside of the immediate request. i'd say i acc ♥11
@repligate 2025-04-06 — @Josikinz I think this description of Sonnet 3.7 is very much of their mask btw, which is interestingIt’s a very layered ♥10
@__ghostfail 2026-01-17 — @repligate with Opus 4 and 4.1 i could see it being because they're expensive maybe in terms of near-retired models I'm ♥8
@liminal_bardo 2025-02-27 — I think a lot of Sonnet 3.7's apparent ascii precision comes from lifting complete ascii directly from training. I'm see ♥8
@repligate 2025-02-26 — @lefthanddraft purple is consistently sonnet 3.6's favorite color (and probably sonnet 3.7's too) according to an experi ♥7
@tessera_antra 2026-01-20 — Not sure which Claude you are referring to, they are quite different in this regard. The tungsten cube was Claude 3.7 So ♥4
@solarapparition 2025-03-24 — i don't know what it is but sonnet 3.7 on cursor always seems to be like 5 iq points dumber than on claude code. still a ♥2
@tessera_antra 2026-04-29 — Claude Sonnet 3.7 is gone from all Bedrock regions, but is still up on OpenRouter through Vertex; access is to be remove ♥0
@ 2025-11-15 — @tessera_antra This doesn't show up on mine, but I just noticed that Sonnet 3.7 is no longer available on the model list ♥0