A developer benchmarked nine LLMs on Spain's VeriFactu tamper-proof invoice ledger, finding that seven of nine models produced the exact SHA-256 hash input text for all 27 test records, but that hones…
A solo developer in Vietnam built Bounty Triage, a 40-case benchmark on Kaggle Benchmarks that tests whether AI models can catch hidden requirements in paid-work listings, and ran 14 models from Anthr…
A developer built Clean Air Window, a Streamlit app that uses Google's open-weight Gemma 4 31B model through OpenRouter to pick the cleanest hour for outdoor activity in smog-prone cities. The tool-ca…
Inception released Mercury Voice, a diffusion LLM tuned for voice agents, to general availability for enterprise customers, claiming a median (p50) time to first answer token of under 320 milliseconds…
A developer built a 150-question benchmark from a 7.2M-page web index to measure frontier models' raw parametric knowledge of 2025-2026 events occurring after their training cutoffs. Across nine model…
A developer has released PaulusAI, an open-source framework for building AI agents that adds a mood system to long-term memory, skill acquisition, and autonomous tool use, distinguishing it from exist…
Channel3 cut AI cost per product by 10× by moving latency-tolerant stages of its product graph pipeline to asynchronous inference on Sail, the company and Sail reported. Most jobs run on Gemma 4 31B t…
A developer built "Blog vs Bytecode," a Kaggle benchmark of 28 data-science code snippets paired with blog-style claims, half of which hide real methodological errors such as scaler leakage or shuffle…
Cohere released North Small Translate, an open-weights mixture-of-experts machine translation model that scores 83.6 on the WMT26 All Languages benchmark, which Cohere says outperforms DeepL NextGen (…
Cohere released North Small Translate, an open-weight sparse Mixture-of-Experts translation model with 218B total and 25B active parameters covering 50 languages, which scores 83.6 on Cohere's WMT26 a…
In a blog post, the team at an AI infrastructure company detailed their optimization of Gemma 4 31B on Google's TPU v6e, boosting prefill efficiency from ~32% to ~63% MFU. They highlighted the chip's …
AgentScript (ASL), an open-source, statically typed language using single-pass S-expressions that compiles to native Rust, Go, TypeScript, and WebAssembly, reduces syntax repair waste in coding agents…
Intel Gaudi 3's vLLM-Gaudi PagedAttention implementation underutilizes hardware for sliding window attention, but optimizations for serving Gemma 4 31B expanded usable KV cache by 3.5×, raised decode …
Nvidia and Cerebras are touting benchmark results at Hot Chips that show inference speeds of 3,400 tokens per second on Gemma 4 31B, but The Register argues these figures are unattainable in real prod…
Duck.ai offers a privacy-focused interface for accessing multiple AI models, including GPT-5.6 Luna, GPT-5.4 mini, Claude Haiku 4.5, Mistral Small 4, gpt-oss-120b, and Gemma 4 31B, without requiring a…
Nvidia has moved its Groq 3 LPX inference chip into full production, claiming 3,400 tokens per second on Gemma 4 31B, which it says is four times faster than Cerebras. However, Nvidia requires at leas…
NVIDIA announced the production release of the NVIDIA Groq 3 LPX, a dedicated interactive inference accelerator for the Vera Rubin NVL72 platform, achieving 3,400 tokens per second on the Gemma 4 31B …
NVIDIA has moved its Groq 3 LPX inference accelerator into full production, targeting the token demands of agentic AI, with Nebius as the first AI cloud to adopt it via its Token Factory service. The …
NVIDIA announced that its NVIDIA Groq 3 LPX rack-scale system is now in full production, delivering 3,400 output tokens per second on a Gemma 4 31B benchmark for 100,000-token long-context use cases, …
Nvidia Corp. announced that its Groq 3 LPX inference accelerator has entered full production, designed to supercharge AI agents by delivering ultra-fast token generation. The chip, unveiled at Hot Chi…