Synthszr Charts — die großen AI-Marken im Wettkampf ums Podium

Llama 3.2

Meta · v3.2 · od 2024-09-25 · 28× · naposledy 06. 9. 2026

Momentum

Meta Llama 3.2 is an open-source model family released on September 25, 2024, comprising four variants: lightweight text-only models at 1B and 3B parameters (designed for edge/mobile deployment) and multimodal vision models at 11B and 90B parameters capable of processing both text and images. All models support a context window of up to 128,000 tokens and use Grouped-Query Attention (GQA). Weights are freely downloadable under the Llama 3.2 Community License; commercial API access is available via third-party providers such as AWS, Azure, and Together AI.

Vývoj momenta

09.07.07.10.

Vlastnosti

Key Benchmark (%)	MMLU (0-shot, CoT): 86.0% (90B model); MMLU 5-shot: 63.4% (3B model); DocVQA: 90.1 (90B Vision); ChartQA: on par with GPT-4o (90B Vision)
Context Window (Tokens)	128,000 tokens (all model sizes: 1B, 3B, 11B, 90B)
License	Llama 3.2 Community License (custom, commercially usable; separate Meta approval required for >700 million monthly active users)
Multimodality	Yes (partial): 11B and 90B Vision models support text and image input (image reasoning, document analysis, captioning, visual grounding); 1B and 3B models are text-only
Price per 1M Tokens	3B Instruct: from $0.030 input / $0.050 output (cheapest provider); 11B Vision Instruct: from $0.060 input / $0.060 output; 3B (Base): from $0.100 input / $0.100 output – prices vary by provider
Release Date	September 25, 2024

Další produkty v této kategorii: Open-source jazykové modely

Zdroje (28)

Company Analysis: Meta

As of 06/10/2026

HOLDSynthszr Vote

Data supports a high-quality core business (Q2 2026 operating margin 31%, EPS $6.18; management expects 2026 operating income above 2025) alongside unusually high capital intensity (2026 capex $130B–$145B). With trailing P/E ~28x and consensus targets implying only moderate upside, the risk/reward is balanced until there is clearer evidence that AI infrastructure spend is translating into sustained incremental revenue and stable free cash flow after depreciation ramps, while regulatory outcomes in the EU/US remain a meaningful swing factor.

Summary

Meta Platforms’ business model is primarily advertising-driven across its Family of Apps (Facebook, Instagram, WhatsApp, Messenger), complemented by emerging monetization in messaging/business tools and longer-duration bets in Reality Labs (AR/VR) and AI products. Core competencies include large-scale consumer product distribution, ad targeting/measurement, recommendation systems, and increasingly AI infrastructure and model deployment at global scale. Market position remains top-tier in global digital advertising, with competitive advantages from multi-app reach, engagement, and a mature performance ads stack. Meta’s scale in data, creator ecosystems, and cross-app monetization (especially Instagram and messaging) supports resilient demand capture versus smaller platforms. Key competitive pressure points remain TikTok/short-form video engagement, Apple/Android platform policy constraints, and regulatory scrutiny in the EU/US. In the most recent reported quarter (Q2 2026, results released July 29, 2026), Meta reported operating margin of 31% and diluted EPS of $6.18, with constant-currency revenue growth cited at 27% year-over-year. Management reiterated that 2026 operating income is expected to exceed 2025 operating income, while also raising the lower end of the expense outlook to incorporate $2.4B of legal-related charges recognized in Q2. Guidance also highlighted FX as an approximately 1% headwind to year-over-year total revenue growth based on then-current exchange rates. These items matter because they frame near-term profitability versus the company’s accelerated AI investment cycle. The dominant near-term financial debate is capital intensity: Meta narrowed its 2026 capital expenditures outlook (including principal payments on finance leases) to $130B–$145B, reflecting aggressive AI infrastructure buildout. This supports product and ads performance improvements but increases execution risk around returns on invested capital and free cash flow volatility. Valuation snapshot (as of Oct 6, 2026): META trades at ~27.95x trailing P/E with a market cap of ~$1.90T and last price ~$741.90 (USD). Converting to EUR requires an FX rate; using an illustrative EURUSD of 1.10 implies ~€674.45 per share (illustrative only). Short- to medium-term outlook is driven by (1) ad demand and pricing, (2) AI-driven engagement/ads efficiency, (3) capex discipline and depreciation trajectory, and (4) regulatory outcomes (notably EU DMA/DSA-related matters and WhatsApp API access requirements).

Key Takeaways

Q2 2026 profitability remained strong (31% operating margin; $6.18 diluted EPS), supporting the view that core ads economics are intact despite heavy investment.
Meta’s 2026 capex plan is exceptionally large ($130B–$145B), making infrastructure execution and ROI the central driver of medium-term equity outcomes.
Management expects 2026 operating income to exceed 2025 operating income, but legal charges ($2.4B recognized in Q2) and higher expense floors add near-term noise.
Regulatory exposure in the EU remains material (DMA/DSA and competition actions), including WhatsApp-related interim measures and ongoing investigations.
Street sentiment remains constructive: multiple aggregators show “Strong Buy” consensus and average price targets around ~$790–$794, implying modest upside versus early-October levels.

Action Ideas

BUY

For investors prioritizing quality large-cap cash generation with AI-driven product leverage: Q2 2026 results show strong profitability (31% operating margin) and management expects 2026 operating income above 2025. If AI-driven ranking/ads tooling continues to lift advertiser ROI, Meta can sustain above-market revenue growth while absorbing higher depreciation over time. Consensus targets around ~$790–$794 suggest the market still underwrites incremental upside if capex translates into durable ad performance and new AI monetization surfaces.

Horizon: 18 mo.

HOLD

For investors already positioned: maintain exposure while monitoring capex and expense trajectory. The investment case is balanced between strong core margins and unusually high infrastructure spend. Holding is supported by continued profitability and constructive analyst consensus, but incremental multiple expansion may be limited until investors gain clearer evidence that AI infrastructure spending is translating into sustained revenue acceleration and stable free cash flow after depreciation ramps.

Horizon: 9 mo.

SELL

For investors with low tolerance for capital intensity and regulatory uncertainty: reduce exposure if portfolio constraints require lower earnings/FCF volatility. The 2026 capex range ($130B–$145B) is large enough that even with strong operating income, free cash flow outcomes can be more variable. If regulatory actions (EU DMA/DSA or competition remedies) tighten monetization or data signals, the risk-adjusted return profile may deteriorate versus other mega-cap alternatives with lower incremental capex needs.

Horizon: 6 mo.

Contrarian Insights

• Consensus optimism may underweight the possibility that AI infrastructure spending across hyperscalers (including Meta) outpaces near-term monetization, creating a longer payback period and higher sensitivity to cost of capital than equity markets currently price.
• EU regulatory outcomes may be a more direct earnings driver than commonly modeled: remedies affecting WhatsApp API access, consent models, or product design could influence engagement and ad performance in Europe beyond one-time fines, with second-order effects on advertiser ROI.