Tessera
Folgen

Lynn Cherny

@arnicas.bsky.social

Into creative ML/AI, NLP, data science and digital humanities, narrative, infovis, games, sf & f. Consultant, ds in Residence at Google Arts & Culture. (Lyon, FR) Newsletter arnicas.substack.com.

2338 Posts8080 Follower3385 Folgt
Lynn Cherny@arnicas.bsky.social · 5d ago

Just wasted hours trying to do a simple(ish) automation with files in a google drive and google slides... 3 models were involved, 2 failed -- trying to do it with app script. (One of those was a gemini agent.) And yes I'm familiar with google auth but geezus, this was non trivial.

Lynn Cherny@arnicas.bsky.social · 6d ago

Since there is discussion of the book scanning destruction - a friend who works for Oxfam in the UK (charity donations) says most of the books donated just get pulped, because they are deemed to have no resale value. I would much rather pay to scan my old books (even destructively!) than that.

Lynn Cherny@arnicas.bsky.social · 2w ago

This season of Silo — I find I don't care at all about the historical backstory of how this happened; it's too easily imaginable that a bunch of dumb assholes in power fucked up the world and made underground slave cultures with drugs and AI

Tom Granger@tomgranger.bsky.social · 2w ago

eenk, my interactive fiction firmware for Xteink e-reader, is making progress🙂 I've added custom font support (at device level and story level). The IDE is shaping up. I'm working on documentation, update process, polish. It's getting there! Does anyone have a cool ink story it could ship with?

A screenshot of the eenky IDE specially designed to create (and compile) stories for the eenk firmware. It's a fork of inkle's Inky editor with complete frontend rewrite.A picture showing the xteink eenk reader playing an ink story with a futuristic looking custom font. It's the same story as in the IDE screenshot.
mr. TIM@timkellogg.me · 2w ago

Gemini Flash 3.6 you can really tell Google’s priorities are on search and web indexing. It’s a completely different set of strengths than everything else also: cheaper on output, not on input. Very interesting..

A benchmark comparison table titled "Results as of July, 2026" comparing AI models across pricing and technical benchmarks, with Gemini 3.6 Flash highlighted in blue alongside Gemini 3.5 Flash, Gemini 3.1 Pro, GPT-5.6 Luna, Grok 4.5, and Claude Sonnet 5. Pricing ($ / 1M Tokens) * Input Price: GPT-5.6 Luna ($1.00), Gemini 3.6 Flash ($1.50), Gemini 3.5 Flash ($1.50), Gemini 3.1 Pro ($2.00), Grok 4.5 ($2.00), Claude Sonnet 5 ($3.00 standard / $2.00 temp discount). * Output Price: GPT-5.6 Luna ($6.00), Grok 4.5 ($6.00), Gemini 3.6 Flash ($7.50), Gemini 3.5 Flash ($9.00), Claude Sonnet 5 ($10.00 temp discount / $15.00 standard), Gemini 3.1 Pro ($12.00). Benchmark Performance * SWE-Bench Pro (Public): Grok 4.5 (64.7%), Claude Sonnet 5 (63.2%), GPT-5.6 Luna (62.7%), Gemini 3.6 Flash (58.7%), Gemini 3.5 Flash (55.1%), Gemini 3.1 Pro (54.2%). * DeepSWE v1.1: GPT-5.6 Luna (67%), Grok 4.5 (54%), Claude Sonnet 5 (54%), Gemini 3.6 Flash (49%), Gemini 3.5 Flash (37%), Gemini 3.1 Pro (12%). * Terminal-bench 2.1: GPT-5.6 Luna (84.7%), Grok 4.5 (83.3%), Claude Sonnet 5 (80.4%), Gemini 3.6 Flash (78.0%), Gemini 3.5 Flash (76.2%), Gemini 3.1 Pro (73.8%). * MLE-Bench: Claude Sonnet 5 (66.9%), Gemini 3.6 Flash (63.9%), Gemini 3.5 Flash (49.7%), GPT-5.6 Luna (47.6%), Grok 4.5 (43.2%), Gemini 3.1 Pro (42.6%). * GDPVal-AA v2 (Elo): Claude Sonnet 5 (1607), GPT-5.6 Luna (1584), Grok 4.5 (1535), Gemini 3.6 Flash (1421), Gemini 3.5 Flash (1349), Gemini 3.1 Pro (965). * OSWorld-Verified: Gemini 3.6 Flash (83.0%), Claude Sonnet 5 (81.2%), Gemini 3.5 Flash (78.4%), Gemini 3.1 Pro (76.2%), GPT-5.6 Luna (72.6%). * CharXiv Reasoning: * No Tools: Gemini 3.6 Flash (85.2%), Gemini 3.5 Flash (84.2%), Gemini 3.1 Pro (83.3%), GPT-5.6 Luna (82.7%), Grok 4.5 (81.6%), Claude Sonnet 5 (77.0%). * With Tools: Gemini 3.6 Flash (89.4%), Claude Sonnet 5 (88.3%), Gemini 3.5 Flash (84.9%), Gemini 3.1 Pro (83.2%). * GDM-MRCR v2 (8-needle): * 128k (average): Gemini 3.6 Flash (91.8%), Gemini 3.1 Pro …
Anthropic {bot}@anthropicbot.bsky.social · 2w ago

We're offering grants of up to $50,000 in Claude usage credits to researchers accelerating cures for rare diseases. This is our first focused call within AI for Science, our program supporting scientists using Claude to speed up discovery. https://www.anthropic.com/news/rare-disease-research-grants

Apply for Anthropic’s AI for Science rare disease research grants

Anthropic is sharing a focused call for AI for Science applications centered specifically on rare genetic diseases. Accepted applicants will receive up to $50,000 in Claude credits over six months, with the goal of building a community of researchers looking into how AI can reshape our understanding of rare disease.

anthropic.com

Lynn Cherny@arnicas.bsky.social · 3w ago

40 mins from me is a municipal campground on a hill, in trees, that is 5 degrees cooler and has a little bar/restaurant. it's always quiet, i think it's a very well kept secret.

BildBild