Tech News
← Home  ·  All topics

Inference

67 GoKawiil briefs on this topic · RSS for inference

Anthropic unveils FlashProve, an open-source tool to verify which AI model generated an output

Anthropic published a blog post detailing FlashProve, an open-source system that provides cryptographic proof confirming which specific AI model produced a given output. The system relies on labs publishing a 'fingerprint' for each released model, which can later be matched to outputs and verified without exposing the model's underlying weights. Anthropic says verification can run locally and could also confirm third-party claims, such as that a cybersecurity scan was performed by a specific frontier model.

AI inference startup Fal valued at $16B after $1.3B funding round

Fal, a San Francisco startup that hosts AI video, image and audio models, raised $1.3 billion in a Series F round led by Bessemer Venture Partners, with Nvidia, Sound Ventures and T. Rowe Price also participating. The deal doubles Fal's valuation to $16 billion from $8 billion in April, and combined with its earlier Sequoia-led Series E, the company has raised $1.65 billion across both rounds. Fal reportedly surpassed $800 million in annualized revenue earlier this year.

Sam Altman says personal computers hoard RAM that should serve cloud AI

OpenAI CEO Sam Altman said in an interview with YouTuber Cheeky Pint that it is 'quite terrible' for people to keep large amounts of DRAM sitting idle in personal computers, arguing memory should be maximized in the cloud given the current supply shortage. He made the comment while discussing local AI inference, where models run directly on a user's device rather than in data centers.

Darkbloom warns Mac compute-sharing users of falling token prices cutting earnings

Darkbloom, a network that lets Mac owners rent out idle compute for AI inference, told members that falling AI token prices will shrink their earnings. The company cited data showing Anthropic's price per million tokens fell 77% between May and October 2026, driven by cheaper, more efficient data center hardware. Darkbloom estimates top earners could see monthly income drop from around $160 to roughly $37 as a result.

Sam Altman calls home-computer DRAM usage 'quite terrible' for AI efficiency

In a podcast interview with Cheeky Pint, OpenAI CEO Sam Altman argued that DRAM sitting idle in personal computers is a poor use of a globally limited resource, since it's only utilized a small fraction of the time. He said that in a supply-constrained environment, memory should instead be concentrated in cloud data centers for maximum utilization, pushing back against the idea that people should run AI models locally.

Open-source spinf-serve lets developers run bulk AI scoring on a single GPU

A developer has released spinf-serve, an Apache 2.0-licensed inference server that reads a piece of content once and answers many fixed-choice questions about it in a single pass, rather than generating text. It runs on a single 24GB NVIDIA GPU, offers an OpenAI-compatible API matching the existing hosted spinf service, and supports both open instruction models and dedicated decision models.

Sam Altman argues cloud RAM beats local RAM amid DRAM shortage

OpenAI CEO Sam Altman said that keeping DRAM concentrated in data centers is more efficient than letting individuals run local inference on personal computers. He argued that memory sitting idle in personal machines most of the time represents poor use of a globally constrained resource, while cloud-based memory can be utilized at much higher rates.

Sam Altman calls home PC RAM usage 'quite terrible' versus data center allocation

OpenAI CEO Sam Altman said in an interview with YouTube channel Cheeky Pint that having DRAM sit idle in personal computers most of the time is a poor use of the world's limited memory supply. He argued that cloud data centers achieve far higher utilization of RAM than home machines running local AI inference 5% of the time.

Tom's Hardware publishes glossary explaining LLM terms like weights, KV cache and quantization

Tom's Hardware has released an explainer breaking down core terminology behind large language models, covering concepts such as models, tokens, weights, KV cache, quantization and inference. The guide is aimed at readers following the growth of local AI hardware, such as Nvidia's RTX Spark, and is intended to support the publication's ongoing AI coverage.

Altman says local AI use wastes scarce memory chips, favors cloud computing

OpenAI CEO Sam Altman said on the Cheeky Pint podcast that people running AI models locally on personal computers is an inefficient use of the world's limited DRAM supply. He argued that cloud-based inference through hyperscalers achieves better utilization than individual users hoarding memory chips for home use. His remarks came as Intel shares fell nearly 30 percent Friday amid a PC market slowdown that analysts linked partly to AI-driven component shortages.

Runware plans assembly-line factory to mass-produce AI data center pods

Romania-based Runware says it is building a second factory to manufacture its modular Sonic Inference Pod data centers on a production line, similar to automobile manufacturing. CEO Flaviu Radulescu told TechCrunch the company aims to scale to 30 megawatts of data center capacity per day by 2028, with pods installable on customer sites within 48 hours.

Nvidia joins $14M Series B funding round for Indian AI startup Simplismart

Simplismart, a four-year-old Bengaluru-based startup, raised $14 million in a Series B round led by Dallas Venture Capital, with participation from Nvidia, Accel, Shastra VC, and Micromax. The all-primary round values the company at roughly $100 million and will support its software that helps businesses run AI models more efficiently on existing computing infrastructure.