BBrainOutput
NVIDIA·Datacenter acceleratorAnnounced

NVIDIA B300 (Blackwell Ultra): Specs & Local-AI Compatibility

Next-step Blackwell Ultra accelerator — placeholder, verify before use.

Some details here are provisional (Announced). Treat specs as approximate and verify against the manufacturer before relying on them or purchasing.

Specs

Memory
288 GB
Memory type
HBM3e
Bandwidth
to verify
Approx FP16
to verify
Architecture
Blackwell Ultra
Process
TSMC 4NP
Power
to verify
Launch
2025

Models this chip can run

Open models graded for a single NVIDIA B300 (Blackwell Ultra), best fit first.

  • Llama 3.1 405B
    Llama · ~405B · 128K ctx · Llama Community License

    Fits at Q4_K_M (~230GB) with ~23.4GB headroom — about 1 concurrent instance.

    Q4_K_M · ~230GBRuns well
  • Qwen3 235B-A22B (MoE)
    Qwen · ~235B · 128K ctx · Apache-2.0

    Fits at Q8_0 (~235GB) with ~18.4GB headroom — about 1 concurrent instance.

    Q8_0 · ~235GBRuns well
  • Qwen2.5 72B
    Qwen · ~72B · 128K ctx · Qwen License

    Fits at FP16 (~145GB) with ~108.4GB headroom — about 1 concurrent instance.

    FP16 · ~145GBRuns well
  • Llama 3.1 70B
    Llama · ~70B · 128K ctx · Llama Community License

    Fits at FP16 (~140GB) with ~113.4GB headroom — about 1 concurrent instance.

    FP16 · ~140GBRuns well
  • Llama 3.3 70B
    Llama · ~70B · 128K ctx · Llama Community License

    Fits at FP16 (~140GB) with ~113.4GB headroom — about 1 concurrent instance.

    FP16 · ~140GBRuns well
  • DeepSeek-R1 Distill Llama 70B
    DeepSeek · ~70B · 128K ctx · MIT

    Fits at FP16 (~140GB) with ~113.4GB headroom — about 1 concurrent instance.

    FP16 · ~140GBRuns well
  • Mixtral 8x7B (MoE)
    Mistral · ~47B · 32K ctx · Apache-2.0

    Fits at FP16 (~90GB) with ~163.4GB headroom — about 2 concurrent instances.

    FP16 · ~90GBRuns well
  • CodeLlama 34B
    CodeLlama · ~34B · 16K ctx · Llama Community License

    Fits at FP16 (~68GB) with ~185.4GB headroom — about 3 concurrent instances.

    FP16 · ~68GBRuns well

Build a private AI Business OS on NVIDIA B300 (Blackwell Ultra)

Run your own AI agents on hardware you control — private by design, no per-seat data leaving your premises. BrainOutput helps you pick the right machine and turn it into a working AI Business OS.

Explore the AI Business OS