Skip to content
Meta· NewsAvailable

Meta Launches Llama 4 With Apple – Flaws Detected in Code Tests

Meta, in partnership with Apple, has launched Llama 4. Despite its MoE architecture and a massive context window, early tests reveal unexpected deficiencies in coding performance compared to Chinese rivals.

By the Aheadline editorial team·19 okt. 2026·2 min read·Source: Entity-watch: Meta AIVerifierad signalAI-generated
Meta Launches Llama 4 With Apple – Flaws Detected in Code Tests
Meta Launches Llama 4 With Apple – Flaws Detected in Code Tests
Meta Launches Llama 4 With Apple – Flaws Detected in Code Tests
By · Policy- & EU-reporter
Last updated
Vad betyder det för mig?

What happened?

Meta has collaborated with Apple to launch the Llama 4 series, featuring the Scout and Maverick models based on a Mixture of Experts (MoE) architecture. Scout contains 109 billion parameters with 17 billion active and supports a context window of 10 million tokens that can run on a single graphics card. Maverick features 400 billion total parameters and 128 experts, but independent tests such as LiveCodeBench indicate that these models perform worse on coding tasks than competitors like Zhipu GLM and Moonshot AI's Kimi.

Key facts

Lanseringsdatum19 oktober 2026
Scout parametrar109 miljarder totalt, 17 miljarder aktiva
Scout kontextfönster10 miljoner tokens
Maverick parametrar400 miljarder totalt, 128 experter
Kommande modellBehemoth (~2 biljoner parametrar)

Why it matters

The launch marks a technical milestone due to the massive context window and the partnership between Meta and Apple. At the same time, the coding results demonstrate that leadership in open source is being challenged by Chinese developers who are performing more strongly in specialized programming despite Meta’s superior computational resources.

Who is affected?

The models are targeted at software developers and companies building AI applications locally or in the cloud. However, developers relying on automated code generation may need to supplement these tools with specialized coding models.

Impact on the EU

As an open model, Llama 4 falls under the EU AI Act, which imposes requirements regarding transparency of training data and technical documentation for general-purpose AI models. Simultaneously, access to model weights enables European companies to operate the system within EU borders.

What else you should know

The reception has also been marked by debate regarding evaluation methods following reports of discrepancies between test versions on LMArena and the final production release. Meta has further announced Behemoth, a future variant with approximately two trillion parameters.

Frequently asked questions

Quick answers about this story

Vad har hänt?
Meta och Apple har lanserat Llama 4-serien med MoE-modellerna Scout och Maverick, som erbjuder upp till 10 miljoner tokens i kontext men presterar svagare än väntat på kodning.
När hände det?
Lanseringen och de första oberoende testerna rapporterades den 19 oktober 2026.
Varför spelar det roll?
Modellerna sätter nya rekord i kontextfönster för öppen källkod men visar att amerikanska AI-jättar möter hård konkurrens inom programmeringsförmåga från kinesiska laboratorier.
Hur påverkas EU och regelverken?
Llama 4 faller under kraven för allmänna AI-modeller (GPAI) i EU:s AI Act, vilket kräver teknisk dokumentation och efterlevnad av upphovsrättsliga regler.
Original source
Entity-watch: Meta AI·tokendun.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

#Meta
[ 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 Launches Llama 4 With Apple – Flaws Detected in Code Te"