Engineering ready
Ministral 3 8B Instruct 2512 FP8 Production Package
Mistral / Ministral 3 8B Instruct 2512 · fp8 · Engineering status: Engineering ready
Transformation lineage
Base releaseMinistral 3 8B Instruct 2512
Model engineeringFP8 Quantization, Runtime packaging, Evaluation and benchmarking
Delivery optionsDownload package / Host with FusionPointAI / Download + host
Transparent pricing
- Base model engineeringUSD 17,400
- FP8 QuantizationUSD 5,900
- Runtime packagingUSD 2,800
- Evaluation and benchmarkingUSD 3,600
- Download packageUSD 1,500
- Managed deployment setupUSD 3,200
- One-time engineeringUSD 34,400
Monthly hosting
- 1 x NVIDIA L4 24GBUSD 250
- Monthly hostingUSD 250
Source release
- Model family
- Mistral
- Base release
- Ministral 3 8B Instruct 2512
- Package type
- Quantized
- Source precision
- Not recorded
- Target precision
- FP8
- Engineering status
- Engineering ready
- Orderability status
- Available to order
Deliverable scope
Not recorded
GPU memory fit
Estimated model weights: 8.72 GB
Estimated KV cache: 2 GB
Estimated required VRAM: 17.67 GB
Estimated available VRAM: 24 GB
Benchmark status
Estimate only
No measured benchmark results are recorded for this package.
Delivery limitations
Configured delivery: Download + host
Delivery options: Download package / Host with FusionPointAI / Download + host
Transformation lineage
Source precision: Not recorded
Target precision: FP8
Transformation operations: FP8 Quantization, Runtime packaging, Evaluation and benchmarking
GPU topology
Target GPU: NVIDIA L4 24GB
GPU count: 1
Runtime packaging: transformers