- Advertisement -
- Meta launches Llama 4 API with Groq and Cerebras as partners
- Lama models offer lower costs and faster output than competitors
- Buy cerebras and Groq can help Meta to have a completely AI stack
On Llamacon 2025, Meta Platforms launched LLAMA 4 API and Pay-Per-Use offered access to its models, hosted on Meta’s own hardware and through Inference Partners Groq and CereBra’s.
As The next platform notedThis step turns effectively with Meta into a platform wool provider, without building a complete general cloud such as its rivals.
Llama is open source, widely adopted and grows rapidly. It has now reached 1.2 billion downloads in his family of models. According to Meta Chief Product Officer Chris Cox, Open Source Ai was seen only two years ago as “a dream”. It is now a worldwide movement, driven by the demand for transparency, control and flexibility.
The Maverick and Scout models from Lama 4 offer large cost benefits. As reported by The next platformThey are 21 to 31 times cheaper per million tokens than GPT-4O and also beat Gemini 2.5 Pro over price and benchmark scores.
Although the performance vary per workload, both models show faster external speeds in many scenarios and they are optimized for regular GPU platforms and newer hardware from Groq and CereBra’s.
In the end of his article, Timothy Prickett Morgan said: “In a briefing, we joked with cerebras that it would be fun if meta platforms would just buy this alternative iron and both cerebra’s and groq from the cloud company completely bought the cloud -business, except perhaps sambaning. Clouds.
The idea is very logical. Llama is on its way to becoming the Android of LLMS – open, flexible and everywhere – and Meta could follow the example of Google and possess the hardware pivot. Buying cerebras and groq would completely yield the inference in -house, improve integration and reduce latency.
At the moment, if The next platform Observed, Meta depends on advertisements for 97% of sales. The LLAMA API offers a clear new electricity, one that can be large enough to require individual financial reporting.
Meta is already the owner of the model. Owning the stack can make it a platform.
Maybe you like it too
- Advertisement -