NVIDIA B300
A data-center GPU accelerator based on the NVIDIA Blackwell architecture. It supports FP4 precision and the second-generation Transformer Engine, and features enhanced memory capacity and bandwidth configurations designed for high-throughput enterprise-grade inference workloads.
Timeline (last 90 days)
Scaleway is deploying an NVIDIA B300 cluster in Copenhagen, with a scale of thousands of GPUs.
The MI355X this time matched or even surpassed the B300 in real agentic workloads, indicating that AMD's weakness is shifting from 'the hardware isn't good enough' to 'the software needs improvement.'
LFM2.5-24B-A2B achieves throughput of over 21,000 tokens/second on a single NVIDIA B300 private cloud accelerator.
LiquidAI's LFM2.5-24B-A2B uses sparse MoE to activate only 2B per token out of 24B total parameters, extends context from 32K to 131K, runs at 184 tok/s on an Apple M5 Max, and reaches aggregate throughput of up to 21K tok/s on a single NVIDIA B300.
On 8 NVIDIA B300 GPUs, a 47-billion-parameter MoE model reaches about 52,000 tokens per second per GPU on the new Olmo-core 3 framework, whereas the old implementation reached only about 19,400, a throughput improvement of about 2.7 times; Ai2 labels this as a "preliminary result."
Together AI announced a partnership with IBM Cloud and NVIDIA to launch a dedicated inference cluster based on NVIDIA B300.
The B300 variant features enhanced memory capacity and bandwidth configurations, making it suitable for running ultra-large models at the 700B parameter scale.
The dedicated inference cluster is built by Together AI, deployed on IBM Cloud, and based on NVIDIA B300.
All Facts (6)
The MI355X this time matched or even surpassed the B300 in real agentic workloads, indicating that AMD's weakness is shifting from 'the hardware isn't good enough' to 'the software needs improvement.'
50%UnverifiedLiquidAI's LFM2.5-24B-A2B uses sparse MoE to activate only 2B per token out of 24B total parameters, extends context from 32K to 131K, runs at 184 tok/s on an Apple M5 Max, and reaches aggregate throughput of up to 21K tok/s on a single NVIDIA B300.
50%UnverifiedThe dedicated inference cluster is built by Together AI, deployed on IBM Cloud, and based on NVIDIA B300.
50%UnverifiedThe B300 variant features enhanced memory capacity and bandwidth configurations, making it suitable for running ultra-large models at the 700B parameter scale.
50%UnverifiedScaleway is deploying an NVIDIA B300 cluster in Copenhagen, with a scale of thousands of GPUs.
50%UnverifiedTogether AI announced a partnership with IBM Cloud and NVIDIA to launch a dedicated inference cluster based on NVIDIA B300.
50%