Skip to content
DigitalNeuron

Independent AI journalism

Clear reporting on the changes shaping the AI industry.

News and analysis on models, chips, companies, investment and policy, with the context readers need to judge what matters.

NewsExplainers

Latest news

View all →
Models & research

DeepSeek Releases V4.1-Flash-1017, Opens Search and Fetch APIs to All Developers

DeepSeek released DeepSeek-V4.1-Flash-1017, which the company says improves benchmark performance, speed, and output length over its predecessor. DeepSeek also opened its Search and Fetch APIs to all developers at $0.75 per 1,000 calls, with free use through certain tools until October 31, 2026. The model is open-sourced on Hugging Face and ModelScope.

  • DeepSeek released DeepSeek-V4.1-Flash-1017 on its website, app, and API, describing it as a Pareto improvement in performance and cost over the prior version.
  • The company says the model is optimized for DeepSeek Harness, Tencent CodeBuddy, and WorkBuddy, and recommends upgrading to DeepSeek Harness v0.2.1-rc.1 or later.
2 min read
Models & research

Anthropic Opens Call for Guest Investigators in Life Sciences

Anthropic is inviting principal investigators from academic, nonprofit, or government labs to apply for 8-12 week collaborations using Claude on problems in human biology, molecular biology, drug discovery, and clinical science, at no cost, with plans to publish or open-source resulting outputs.

2 min read
Tools & products

AWS Upgrades Amazon Polly's Generative Engine With 14 New Voices

AWS announced a general-availability upgrade to Amazon Polly's Generative engine, adding 14 new voices across 9 locales, including first-time Hindi and Finnish support, bringing the total to 57 generative voices across 11 languages and 25 locales, with no code or pricing changes for existing users.

2 min read
Tools & products

Call of Duty: Modern Warfare 4 Launches on GeForce NOW

NVIDIA announced that Call of Duty: Modern Warfare 4 launched on GeForce NOW, with Ultimate members getting RTX 5080-class ray tracing. Fire TV Stick users can now buy memberships via Amazon, and FINAL FANTASY RESONANCE and an Elder Scrolls Online season also joined the cloud library.

2 min read

AI news briefing

Headlines from other outlets, collected automatically and updated daily. Every item links to the original.

View all →

Analysis and explainers

The technology and business of AI, explained from the ground up.

View all →
Tools & products

How to choose an AI model for a real product

Start from the constraints rather than the leaderboard. Write down your latency budget, your cost ceiling per request, your context requirement, whether you need tool calling or structured output, and where the data is allowed to go. Those five usually eliminate most candidates. Test the remaining two or three on fifty real cases from your own traffic, compare per-case rather than on average, and pick the cheapest one that passes.

5 min read
Models & research

Fine-tuning, retrieval or a better prompt: how to choose

Diagnose the failure first. If the model does not know something, that is a knowledge gap and retrieval fixes it. If the model knows but answers in the wrong shape, tone or format, that is a behaviour gap and prompting fixes it first, fine-tuning second. If the model fails at a specialised skill after both, fine-tuning is the remaining option. Prompting is cheapest and reversible, retrieval is the right default for facts, and fine-tuning is the most expensive and least reversible of the three.

5 min read
Chips & infrastructure

Why does AI use so much electricity?

AI accelerators draw far more power per rack than traditional servers, and that power has to be delivered, cooled and paid for continuously. Training a large model is a one-off spike; serving it to millions of users is a permanent load, and inference is what dominates energy use over a deployed model's life.

4 min read

Practical AI guides

Useful guidance based on official documentation and hands-on practice.

View all →
Tools & products

When a Claude Project or a ChatGPT Project actually earns its keep, per each company's own help pages

Both companies frame Projects the same way: a workspace that holds its own chat history, uploaded reference files, and standing instructions, so you stop re-explaining context every session. Claude caps free accounts at 5 projects and scales knowledge capacity up to 10x on paid plans; ChatGPT scopes a project's memory to chats inside that project only, not across your whole account.

3 min read

Browse by topic

View all →
Models & researchNew frontier models, benchmarks, training methods and the research behind them.Agents & automationTool-using AI systems, autonomous workflows and what they actually automate.Chips & infrastructureAccelerators, data centres, memory supply and the power bill behind AI.Business & fundingRounds, revenue, pricing and the economics of shipping AI products.Policy & regulationLaws, enforcement, copyright and the rules AI has to live inside.Safety & ethicsEvaluations, misuse, bias, alignment research and incident reporting.Open source AIOpen-weight models, permissive licences and the community stack around them.Tools & productsApplications people can use today, with honest notes on limits and cost.

AI glossary

Short, precise definitions of the terms you keep seeing in headlines.

18 articles →

Model directory

A side-by-side reference of the frontier models people are actually deploying.

View all →

The five-minute AI briefing

One email a week. What shipped, what it costs, and what it means — no hype.

No spam. Unsubscribe in one click.