← All tags

Articles: Open Weights

16 articles on this topic.

LLM translation: OVHcloud's Olala tested against DeepLAI

LLM translation: OVHcloud's Olala tested against DeepL

LLM translation tested: OVHcloud's Olala Translate against DeepL and gpt-oss-120b on 100 business document chunks. Fidelity, fluency, cost and speed compared.

10 min
Qwen 4 27B vs Qwen 3.8 27B: What to Expect from Qwen 4AI

Qwen 4 27B vs Qwen 3.8 27B: What to Expect from Qwen 4

Qwen4-Max shipped on November 16 without weights; Qwen 4 27B is still not out. What is confirmed, and how Qwen 3.8 27B compares with Mistral models.

16 min
Updated on
Olala mini, OVH's LLM: its benchmarks and what it's worthAI

Olala mini, OVH's LLM: its benchmarks and what it's worth

Olala mini 7A1B, OVH's first LLM, is now on Hugging Face. Its scores against Qwen and Ministral, how to run it locally, what it can do for businesses.

11 min
Updated on
Local LLM Server: What Hardware Does a Company Need?AI

Local LLM Server: What Hardware Does a Company Need?

Local LLM server: workstation, RTX PRO 6000 card or H200. What each tier serves with Qwen 3.8 27B, measured throughput and prices checked in November 2026.

13 min
Updated on
Open-source AI in the enterprise: licenses and real savingsConcepts

Open-source AI in the enterprise: licenses and real savings

Open-source AI in the enterprise: what each open model license really allows, how much it saves compared with closed APIs, and why the rules are tightening.

10 min
Updated on
DeepSeek V4.1 Pro vs DeepSeek V4 Pro: What to ExpectAI

DeepSeek V4.1 Pro vs DeepSeek V4 Pro: What to Expect

DeepSeek V4.1 Pro is not out as of November 16. What is confirmed, what the leaks are worth, and DeepSeek V4 Pro today: pricing, MIT license, data location.

9 min
LLM gateway: OpenRouter, LiteLLM or build your own?Guide

LLM gateway: OpenRouter, LiteLLM or build your own?

LLM gateway or AI gateway: what it does, OpenRouter vs LiteLLM, what it costs and where your data goes. Our own price and speed readings included.

10 min
Updated on
Open-weight vs open-source: what an open-weight model really isConcepts

Open-weight vs open-source: what an open-weight model really is

Open-weight or open-source model: definition, differences, licences and enterprise use. Why an LLM is almost never open-source the way Linux is.

6 min
Updated on
Pick a number between 1 and 10: why AI always says 7Concepts

Pick a number between 1 and 10: why AI always says 7

We asked seven AI models for a number between 1 and 10, 14,000 times. Five answer 7 almost every time, whatever the temperature. Where does this bias come from?

9 min
Mixture of Experts (MoE) vs Dense Models: A Simple GuideConcepts

Mixture of Experts (MoE) vs Dense Models: A Simple Guide

Mixture of experts (MoE) vs dense LLMs: how each architecture works, what it changes for your GPUs, speed and price per token, and which one to choose.

AI self-replication: Anthropic's study on AI wormsConcepts

AI self-replication: Anthropic's study on AI worms

Can an AI replicate itself? What Anthropic tested on connected devices, AI worm versus AI virus, the study's limits and what businesses can do about it today.

11 min
AI Distillation: Everyone Blames China, Everyone Does ItConcepts

AI Distillation: Everyone Blames China, Everyone Does It

OpenAI and Anthropic accuse Chinese labs of distilling their models. Yet Meta, Google, Mistral and even Musk's xAI distill too. Here is what we actually know.

11 min
AI Inference: the 7 ways to run your models in the enterpriseGuide

AI Inference: the 7 ways to run your models in the enterprise

Enterprise AI inference: 7 approaches compared, from buying GPUs to a turnkey provider. Real costs, sovereignty, latency and how to choose.

11 min
Updated on
Enterprise AI SLA: one AI provider for your agents is not enoughGuide

Enterprise AI SLA: one AI provider for your agents is not enough

Why one provider is not enough for an enterprise AI SLA. Compare commitments, multi-provider gateways and fallbacks for RAG and AI assistants.

3 min
Updated on
Mistral OCR: cut your bill 10x with open weightsGuide

Mistral OCR: cut your bill 10x with open weights

Mistral OCR costs $4 per 1,000 pages. An open-weight OCR on a European GPU drops to $0.15 and stays sovereign. Benchmark, break-even point and pitfalls.

10 min
How a Failed Demo Forced Us to Rethink Our Embedding StrategyConcepts

How a Failed Demo Forced Us to Rethink Our Embedding Strategy

Slow embedding APIs and vendor lock-in: our production lessons, and why ATG now calls Qwen 3 Embedding 8B on several providers at once.

4 min