Skip to content
Lanseringar· NewsAvailable

Meta Releases Llama 3.2: New Open Models for Local Execution

Meta has launched Llama 3.2, a new series of open language models that introduce smaller model sizes for local execution and improved performance across benchmarks.

By the Aheadline editorial team·18 sep. 2026·2 min read·Source: Entity-watch: Meta AIVerifierad signalAI-generated
Meta Releases Llama 3.2: New Open Models for Local Execution
Meta Releases Llama 3.2: New Open Models for Local Execution
Meta Releases Llama 3.2: New Open Models for Local Execution
By · Policy- & EU-reporter

What happened?

Meta has released Llama 3.2, a new generation of its open language models. The new series includes smaller models with 1B and 3B parameters designed to run locally on devices, in addition to larger variants. In evaluations on the GLUE benchmark, Llama 3.2 achieves a score of 82.5 points, an increase from its predecessor's 81.5 points. The models support context lengths of up to 32K tokens, and the source code has been made publicly available.

Key facts

ModellserieLlama 3.2
GLUE-resultat Llama 3.282,5 poäng
GLUE-resultat Llama 3.181,5 poäng
Minsta modellstorlekar1B och 3B parametrar
Maximal kontextlängd32K tokens

Why it matters

The new model sizes make it easier to run generative AI directly on consumer hardware and mobile devices without the need for significant cloud resources. By offering open weights and an expanded context window, Meta continues to challenge closed-source model alternatives in the market.

Who is affected?

The launch primarily concerns AI developers, researchers, and companies that want to run language models locally or build applications using open weights. End users of applications that integrate Meta models on mobile devices are also affected by the more efficient and smaller model sizes.

Impact on the EU

Llama 3.2 has been released as open models globally, making them accessible to developers within the EU for local execution or distribution via cloud services. Because the models are also available in smaller sizes such as 1B and 3B, they are well-suited for on-device use in Europe without the need to transfer data to external servers.

What else you should know

In addition to the launch of Llama 3.2, several independent evaluations have compared open models against large-scale alternatives. At the same time, research and development are ongoing regarding how large language models handle internal knowledge retrieval and reward mechanisms during fine-tuning.

Frequently asked questions

Quick answers about this story

Vad har hänt?
Meta har offentliggjort Llama 3.2, en ny generation öppna språkmodeller som inkluderar mindre storlekar som 1B och 3B samt utökat stöd för kontextlängd upp till 32K tokens.
När hände det?
Llama 3.2 lanserades och dokumenterades i september 2024.
Varför spelar det roll?
Modellerna gör det möjligt att köra avancerad AI lokalt på enheter och ger utvecklare tillgång till öppna vikter med förbättrad prestanda på standardiserade testsviter.
Påverkar det EU?
Ja, modellerna är tillgängliga globalt och kan laddas ned för lokal körning eller nås via molnleverantörer i EU.
Original source
Entity-watch: Meta AI·airebao.com

The link opens in a new window and leads to the publisher's own site.

Verifierad signal

Källan har spårats automatiskt från utgivaren via Aheadlines signalkedja.

AI-verktyg i artikeln

Topics

#Open Source#AI-benchmarking#Large Language Models (LLMs)#Models#Meta AI
[ STAY UP TO DATE ]

Get similar news straight to your inbox

No affiliate linksCancel anytimeGDPR-friendly
[ Frequency ]
[ What do you want to read about? ]

You'll receive updates on 2 topics.

The reader's room

Send in a question or an addition. The newsroom reads everything before it's published and replies when relevant. No AI-generated text – just people.

Sign in to submit a comment or question.

Loading comments…
How this affects you

Read the article through your role

  • Decide whether this affects strategy over 6–12 months or is just noise.
  • Discuss with leadership: do we own the right question or does ownership need to move?
  • Ask: what risk are we taking by NOT acting on this this quarter?

Generated angle — not editorial analysis of "Meta Releases Llama 3.2: New Open Models for Local Execution"