Flowers Labs
Synthetic Data
Upload your input data and generate synthetic data for cheap. Tokens wasted due to RLHF rejections are not counted.
ModelEndpointInput (USD)Output (USD)
Mistral 7Bmistral-7b0$ / 1M tokens0.1$ / 1M tokens
Mistral 7B Instructmistral-7b-i0$ / 1M tokens0.1$ / 1M tokens
Mixtral 8x7Bmixtral-8x7b0$ / 1M tokens0.25$ / 1M tokens
Mixtral 8x7B Instructmixtral-8x7b-i0$ / 1M tokens0.25$ / 1M tokens
Flowers Smallflowers-small3$ / 1M tokens10$ / 1M tokens
Flowers Mediumflowers-medium10$ / 1M tokens45$ / 1M tokens
Compute
  • 1320 FP16 TFLOPS
  • CUDA support (Nvidia GPUs yay!)
  • 96 GB 1TB/s GPU memory
  • 28 GB/s NVMe read bandwidth
  • 8 TB NVMe and 256 GB RAM
  • AMD Epyc 32 cores
  • 2x 25 GbE + 2x 10 GbE
  • 10 SATA 6 Gb/s ports
  • $12,000 (shipping Q3 2024)