Model Terminal

fp8_dynamic_moe-e2e

A test/development model artifact published under the nm-testing Hugging Face organization, providing an FP8-quantized mixture-of-experts (MoE) model for end-to-end inference and serving pipeline validation. It is oriented toward benchmarking and validating FP8 dynamic quantization workflows across model, runtime, and deployment layers rather than serving as a production-facing model release. Source: supabase.

Value score
Not scored
Context
tokens
Max output
tokens
Price

Capability radar

Identity

Developer
NM Testing
Openness
Modalities
Release
Knowledge cutoff
Deprecation
API docs

Benchmarks

No benchmark scores yet.

Compare nearby

Pricing

Input / 1M
Output / 1M
Speed