// Open-model VRAM table · family
InternLM / Intern-S: local deployment & VRAM
12 open-weight models from Shanghai AI Lab (28 official repos incl. quantised builds). At 4-bit they range from 2.6 GB (internlm2_5-1_8b-chat) to 669 GB (Intern-S1-Pro).
official orgs: internlmlatest 2026-09-13memory = estimate
01Text LLM (6)
| Model | Params | Context | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|---|
| Atria-Dawn-Preview | 753.3B | 1M | 1.66 TB | 504 GB | 8 × H100 (640GB) | — | Commercial OK | 2026-09-11 |
| internlm3-8b-instruct | 8.8B | 32K | 20.3 GB | 6.8 GB | 8GB GPU | Mac 16GB | Commercial OK | 2025-01-13 |
| internlm2_5-1_8b-chat | 1.89B | 32K | 5.5 GB | 2.6 GB | 8GB GPU | Mac 16GB | Custom licence – read it | 2024-07-30 |
| internlm2_5-7b-chat | 7.74B | 32K | 18.6 GB | 6.7 GB | 8GB GPU | Mac 16GB | Custom licence – read it | 2024-06-27 |
| internlm2-math-plus-7b ↗ | 7.74B | 8K | 18.6 GB | 6.7 GB | 8GB GPU | Mac 16GB | Custom licence – read it | 2024-05-24 |
| internlm2-math-plus-mixtral8x22b | 140.6B | 64K | 312 GB | 96.2 GB | H200 141GB | Mac 256GB | Custom licence – read it | 2024-05-24 |
02Vision-language (6)
| Model | Params | Context | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|---|
| Intern-S2-397B | 403.4B | 256K | 888 GB | 270 GB | 4 × RTX PRO 6000 (384GB) | Mac 512GB | Commercial OK | 2026-09-13 |
| Intern-S2-Mobius ↗ | 36B | 256K | 79.8 GB | 24.7 GB | RTX 5090 32GB | Mac 64GB | Commercial OK | 2026-07-29 |
| Intern-S2-Preview ↗ | 36.1B | 256K | 80.1 GB | 24.7 GB | RTX 5090 32GB | Mac 64GB | Commercial OK | 2026-05-15 |
| Intern-S1-Pro | 1T / 22B | 256K | 2.20 TB | 669 GB | 8 × H200 (1,128GB) | — | Commercial OK | 2026-02-02 |
| Intern-S1-mini | 8.54B | 64K | 20.5 GB | 7.4 GB | 12GB GPU | Mac 16GB | Commercial OK | 2025-08-18 |
| Intern-S1 | 240.7B | 64K | 532 GB | 163 GB | 4 × RTX PRO 6000 (384GB) | Mac 256GB | Commercial OK | 2025-07-24 |
03Other families
Qwen · IBM Granite · NVIDIA Nemotron · Gemma · Mistral / Magistral / Devstral · GLM / CogVideoX (Z.ai) · DeepSeek · Ai2 OLMo / Molmo · Ant Ling / Ring / Ming · Liquid LFM · MiniCPM / VoxCPM · Tencent Hunyuan / Hy · Phi · Llama · Meituan LongCat · Jina embeddings & rerankers · Wan · Kimi · Cohere Command / Aya · Xiaomi MiMo · FLUX · Microsoft VibeVoice / Fara / Harrier · StepFun Step · MiniMax · Baidu ERNIE · NVIDIA Parakeet / Canary speech · Stable Diffusion / Stable Audio · Yi (01.AI) · ByteDance Seed · SmolLM / SmolVLM · BGE embeddings & rerankers · LTX Video · FunAudioLLM speech · gpt-oss · poolside Laguna · Whisper · Baichuan · Thinking Machines Inkling · xAI Grok · rednote dots
Totals include 10% + 0.5GB overhead and one 8K-token KV cache for text models. Estimates only. Method →