Submissions from twitter.com/alexalbert__

		We removed the τ2-bench airline eval because Opus 4.5 was too clever (twitter.com/alexalbert__)
		2 points by stared 42 days ago \| past
		Claude Code: GitHub Actions Integration (twitter.com/alexalbert__)
		1 point by tosh 6 months ago \| past
		LLM-Oriented API Design (twitter.com/alexalbert__)
		3 points by delifue 6 months ago \| past
		A Conversation on Claude Code (twitter.com/alexalbert__)
		3 points by tosh 7 months ago \| past \| 6 comments
		Claude 3.5 Haiku is now available (twitter.com/alexalbert__)
		10 points by wut42 on Nov 4, 2024 \| past
		The Anthropic Message Batches API - 50% savings! (twitter.com/alexalbert__)
		2 points by brianjking on Oct 8, 2024 \| past \| 3 comments
		Anthropic: Claude 3.5 Sonnet output limit doubled from 4096 to 8192 (twitter.com/alexalbert__)
		29 points by tosh on July 16, 2024 \| past \| 1 comment
		Favorite Claude-Generated Artifacts from Anthropic's Slack Channel (twitter.com/alexalbert__)
		1 point by r_singh on June 22, 2024 \| past
		Claude 3 Opus suspects it is being tested from benchmark question (twitter.com/alexalbert__)
		42 points by magnio on March 5, 2024 \| past \| 13 comments
		Internal testing on Claude 3 Opus revealed "interesting behavior" (twitter.com/alexalbert__)
		20 points by behnamoh on March 4, 2024 \| past
		GPT-4 provides little protection against leaking system prompts (twitter.com/alexalbert__)
		4 points by rafiste on April 12, 2023 \| past
		Using prompt compression to create the most complex GPT-4 jailbreak ever made (twitter.com/alexalbert__)
		8 points by rafiste on April 5, 2023 \| past \| 1 comment
		A token-smuggling jailbreak for ChatGPT-4 (twitter.com/alexalbert__)
		444 points by rafiste on March 16, 2023 \| past \| 268 comments