Smallest math models for H100 80GB

Verified open math models that fit in 80 GB on H100 80GB, smallest first — with license clarity and hosted-download status. Top pick: mradermacher/Phi-4-mini-reasoning-heretic-GGUF (~1.87 GB).

Hugging Bay is intentionally crawlable for search engines, answer engines, user-requested AI fetches, frontier AI assistants, and open-source agents. Public pages describe verified open-source AI artifact metadata, provenance, licenses, community trust signals, and selective hosted mirrors with hashes. Hugging Bay does not bypass gated or restricted upstream access controls.

Machine-readable entrypoints: robots.txt, AI search guidance, well-known AI search JSON, AI citation packs, well-known AI citation JSON, llms.txt, llms-full.txt, AI crawler policy, AI bot allowlist, answer-engine manifest, agent manifest, OpenAPI, and indexing status.

Verified open math models that fit in 80 GB on H100 80GB, smallest first — with license clarity and hosted-download status. Top pick: mradermacher/Phi-4-mini-reasoning-heretic-GGUF (~1.87 GB).

#ModelEst. VRAMLicenseCommercialHostingFit page
1mradermacher/Phi-4-mini-reasoning-heretic-GGUF~1.87 GBmityesexternalcheck
2amusktweewt/tiny-model-500M-chat-v2~1.87 GBmityesexternalcheck
3openbmb/DensingLaw-ScalingModels~1.87 GBapache-2.0yesexternalcheck
4huihui-ai/phi-4-abliterated~1.87 GBmityesexternalcheck
5mradermacher/CogitoZ14-GGUF~1.87 GBapache-2.0yesexternalcheck
6bartowski/microsoft_Phi-4-mini-reasoning-GGUF~1.87 GBmityesexternalcheck
7mlx-community/Unsloth-Phi-4-4bit~1.87 GBmityesexternalcheck
8AQ-MedAI/Diver-Retriever-0.6B~3.48 GBapache-2.0yesexternalcheck
9RCDWealth/Dolphin3.0-Qwen2.5-1.5B-Q5_K_M.gguf~3.71 GBapache-2.0yesexternalcheck
10build-small-hackathon/minicpm5-1b-math-lora~4.28 GBapache-2.0yesexternalcheck

Other GPUs: RTX 4090 24GB · RTX 3090 24GB · RTX 3060 12GB · NVIDIA T4 16GB (free Colab) · A100 80GB · mac-m3-max-64

Other tasks on H100 80GB: coding · chat · reasoning · translation · summarization · vision · audio · rag

Fit = weights + KV-cache estimate under 80 GB. Recomputed hourly from live catalog state.