liquid
PulseAugur coverage of liquid — every cluster mentioning liquid across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
Liquid releases LFM3-1.7B open-weight model with 262k context
LFM3-1.7B, an open-weight model from Liquid, has been released with a context window of 262.1k tokens. This model offers significant efficiency for its size, with pricing at $0.08/$0.2 per 1 million tokens. The release …
-
OpenAI releases Decisions API to compete in fast, low-cost intelligence market
OpenAI has released its Decisions API, a model designed for rapid and cost-effective decision-making, entering a competitive market. This new API, based on a reconfigured ChatGPT Luna model, significantly reduces latenc…
-
Developer shares method to find slow Shopify Liquid code
A developer shares a method for identifying performance bottlenecks in Shopify Liquid code, focusing on server-side rendering times. The technique involves using the Shopify Theme Inspector for Chrome to pinpoint slow s…
-
Liquid releases d1 ultra decision model via API
Liquid has released its most advanced model to date, the d1 ultra decision model. This new offering is accessible through an API, allowing users to integrate its capabilities into their applications. The announcement wa…
-
New open-weight model LFM2.5-24B-A2B (Liquid) released with 128k context
A new open-weight model named LFM2.5-24B-A2B, also referred to as Liquid, has been released and added to a leaderboard tracking the latest AI models. This model supports a context window of 128,000 tokens and has associ…
-
New backdoor attack targets multimodal AI models
Researchers have identified a new type of backdoor vulnerability, dubbed ToBAC, that specifically targets Unified Autoregressive Models (UAMs). These models, capable of generating both text and image tokens, are suscept…
-
Tailwind CSS unlikely to die, Shopify's FOSS support cited
Shopify's continued maintenance of open-source projects like Toxiproxy and Bootsnap suggests that Tailwind CSS is unlikely to disappear. The author anticipates that Tailwind CSS will join the ranks of other well-support…
-
Guide to Delegating Tasks to Local LLMs with Benchmarks
A technical article explores the capabilities and limitations of local Large Language Models (LLMs), providing benchmarks to guide users on task delegation. The author, quintetkit, uses the Liquid templating language to…
-
Developer's shared code edit breaks multiple tools
A developer at RAXXO shared an experience where a single line edit in a shared Liquid snippet unexpectedly altered the behavior of two other RAXXO products. This snippet was used across multiple tools for elements like …
-
Learn Prompt Regression Testing for LLMs with a Golden-Set Harness
This article introduces prompt regression testing for LLMs, a method to ensure consistent outputs by comparing new responses against a predefined set of "golden" examples. The author, Hsiang-Cheng Chen, demonstrates how…
-
Open-weights LFM2.5-2.6B model with 128k context added to tracker
The LFM2.5-2.6B model, also known as Liquid, has been added to an AI model tracking service. This model features open weights and a free tier, with a context window of 128,000 tokens.
-
Developer creates custom harness to benchmark coding LLMs on personal bugs
A developer has created a Python-based harness to evaluate coding LLMs against a personal corpus of bugs, rather than relying on public benchmarks like SWE-bench. This approach aims to provide more relevant performance …
-
Sonnet model update causes Liquid syntax errors with JSON output
An update to the Sonnet model has introduced a change where raw JSON output is now enclosed within a markdown fence. This modification has caused issues for users relying on Liquid syntax, specifically triggering an 'Un…
-
Kimi K3, DeepSeek V4, and Mistral models lead open-weight AI surge · 1 source tracked
The open-weight AI ecosystem is experiencing rapid advancement with several new models expected soon. Kimi K3 is anticipated shortly, followed by the general availability of DeepSeek V4, new Liquid models, and upcoming …
-
Shopify Developer Reuses Key Sections to Speed Up Store Builds
The author details a strategy for efficiently building Shopify stores by reusing pre-designed theme sections. They focus on four key sections: a hero, a proof band, an FAQ, and a call-to-action, which are copied into ea…
-
New methods tackle concept erasure in autoregressive image generation · 2 sources
Two new research papers introduce methods for concept erasure in autoregressive image generation models, addressing concerns about misuse and the creation of unsafe content. The first paper, Obliviate, proposes a guidan…
-
Shopify theme development time cut in half with Liquid patterns
A developer significantly reduced Shopify theme build times by implementing five key Liquid patterns. These patterns streamline the process of creating theme sections, particularly by using schema fragments to avoid rep…
-
Cheapest LLM APIs for Startups in 2026: Open-Weights Models Offer Major Savings
For startups in 2026, utilizing open-weights LLM APIs through platforms like OpenRouter offers a significant cost advantage. Models such as Meta's Llama 3.1 8B Instruct and Microsoft's Phi-4 provide substantial savings,…
-
Chat template errors can break prompt caching
This article discusses how specific chat template patterns in programming can inadvertently hinder prompt caching. It highlights that certain syntax, like the 'endraw' tag in Liquid, can cause errors and prevent efficie…
-
Shopify enables AI agent endpoints for e-commerce
Shopify has quietly rolled out agent-facing endpoints across its platform, automatically serving files like /llms.txt and /agents.md to AI agents. These endpoints provide information about store capabilities, supported …