Skip to main content

Blog

Articles, guides, and insights on local artificial intelligence

7 articles found per "open-weight"

Qwen3.8-2.4T-A95B: the math of running 2.4T open-weight parameters, 95B active
ModelsAdvanced

Qwen3.8-2.4T-A95B: the math of running 2.4T open-weight parameters, 95B active

The first Qwen-Max-class open-weight model (2.4T parameters, 95B active) outgrows any consumer node: the math on weights, multi-node hardware, 36-month amortization, and power costs, and when running it locally makes sense.

iaLocale Team
November 22, 2026
Mistral Large 4 in open weights: a 1T MoE (52B active) for on-premise inference
ModelsIntermediate

Mistral Large 4 in open weights: a 1T MoE (52B active) for on-premise inference

Mistral released Large 4 in open weights (1.03T total, 52B active, 1M context, open-weight license): an article teaching how to read MoE specs and do the RAM, quantization and speed math for on-premise inference.

iaLocale Team
November 14, 2026
Enterprise agents with open-weight models: Gartner's cost report, the math and the 90% rule
ComparisonsIntermediate

Enterprise agents with open-weight models: Gartner's cost report, the math and the 90% rule

The Gartner report (November 2026): 90% of an enterprise agent's steps on open-weight models, at one fifth of a frontier's cost. The three key numbers, the assumptions, and when it's really worth it.

iaLocale Team
November 12, 2026
On-device phone agents: what open mobile agents actually bring (after OpenMobile-2)
Use casesIntermediate

On-device phone agents: what open mobile agents actually bring (after OpenMobile-2)

OpenMobile-2 opens up mobile agent training environments and recipes, and LFM3-1.7B shows a 1.7GB agent already runs on your phone: numbers, minimum requirements and honest limits.

iaLocale Team
November 5, 2026
Where local models learn: open weights explained without the fluff
FundamentalsBeginner

Where local models learn: open weights explained without the fluff

The model doesn't hook into the internet for its abilities: it absorbed them once at pre-training and executes them on your hardware. Open weights isn't open data, and the models come from many labs, not one.

iaLocale Team
October 27, 2026
Ling 3.1 Flash: 560B open-weight with MIT license and 1M context — what changes if you have 24GB of VRAM
ModelsIntermediate

Ling 3.1 Flash: 560B open-weight with MIT license and 1M context — what changes if you have 24GB of VRAM

inclusionAI's Ling 3.1 Flash: hybrid MoE 560B total / 25B active, native 1M-token context, FP8 checkpoint, MIT license. What it concretely means for 24GB of VRAM: quantizations, parallelism and API costs.

iaLocale Team
October 19, 2026
Aleph Alpha's Kolibri: the 78B MoE that runs (almost) locally
ModelsIntermediate

Aleph Alpha's Kolibri: the 78B MoE that runs (almost) locally

Aleph Alpha's Kolibri (78.1B/3.46B, Apache 2.0, FP8) is the European open-weight MoE: reasoning, tool calling and context up to 1M tokens. In FP8 it weighs ~78GB of VRAM: it runs on an RTX 5090 32GB + shared RAM or on unified-memory Macs; for practical adoption the kolibri1 architecture in llama.cpp and Ollama/LM Studio builds are still missing.

iaLocale Team
October 18, 2026