Intern S2 397B — Hardware Requirements & GPU Compatibility
VisionFunctions
Intern-S2-397B is Shanghai AI Laboratory's multimodal foundation model, built for scientific intelligence and long-horizon agent tasks. It is a Mixture-of-Experts model with roughly 397 billion parameters, using a vision-language pretraining approach that learns directly from raw pages of scientific literature, jointly modeling text and visual relationships without an intermediate parsing step. It supports a thinking mode for deeper reasoning, enabled by default, and was trained with reinforcement learning across more than 20 scientific domains. It was evaluated with up to a 256K token context for text reasoning and 64K for multimodal inputs, and is released under the Apache 2.0 license. At roughly 397 billion parameters, running it locally needs multi-GPU server-class hardware; most users will use a hosted endpoint instead.
Intern S2 397B runs on a 256 GB+ Mac or a datacenter GPU
Best local pick → Mac Studio (M3 Ultra, 256GB) · Q4_K_M · 242.6 GB · ~27 tok/s
Macs from 128 GB (IQ2_XS · ~40 tok/s)
Apache 2.0
Not set up for it?
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Intern S2 397B
One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.
Recommended
ollama run hf.co/bartowski/Intern-S2-397B-GGUF:Q4_K_MQ4_K_M runs on one Instinct MI325X 256 GB · 16K context
Download: Intern-S2-397B-Q4_K_M/Intern-S2-397B-Q4_K_M-00001-of-00007.gguf254.58 GB + 6 partsbartowski
Ways to runOther ways
each opens step-by-stepRun in cloud
Rent 4× H100 SXM · Q4_K_M
- Estimated hourly price
- ~$10.76/hr
- est., whole server
- GPUs
- 4 GPUs
- 320 GB of VRAM
$5+ bonus for new users
How Much VRAM Does Intern S2 397B Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | Runs on | |
|---|---|---|---|---|
| 171.9 GB | 172.3 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Runs on B200 192 GB | |
| 176.9 GB | 177.3 GB | 3-bit small — 3-bit small quantization | Runs on B200 192 GB | |
| 197.1 GB | 197.5 GB | 3-bit medium — 3-bit medium quantization | Runs on Instinct MI325X 256 GB | |
| 202.1 GB | 202.5 GB | 4-bit · legacy — 4-bit legacy quantization | Runs on Instinct MI325X 256 GB | |
| 242.5 GB | 242.9 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Runs on Instinct MI325X 256 GB | |
| 287.9 GB | 288.2 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Runs on Instinct MI350X 288 GB | |
| 333.3 GB | 333.6 GB | 6-bit, very good — 6-bit quantization, very good quality | Runs on 5× H100 SXM 80 GB | |
| 403.9 GB | 404.2 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Runs on 6× H100 SXM 80 GB |
Selected Q4_K_M
4-bit medium quantization242.6GBat 16K context
- Weights
- 242.1 GB
- KV cache
- 0.3 GB
- Overhead
- 0.3 GB
Server-class: ~242.6 GB at 16K context · one Instinct MI325X 256 GB
Which Hardware Can Run Intern S2 397B?
Q4_K_M · 242.5 GBIntern S2 397B (Q4_K_M) requires 242.5 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 316+ GB is recommended. Using the full 262K context window can add up to 3.9 GB, bringing total usage to 246.4 GB. No consumer GPU has enough memory. 2 devices with unified memory can run Intern S2 397B, including Mac Studio (M3 Ultra, 512GB), Mac Studio (M3 Ultra, 256GB).
128–144 GB
Q4_K_M · best fit IQ2_S (126.6 GB)
4 GPUs
GH200 Grace Hopper Superchip~175Simulate output on GH200 Grace Hopper SuperchipH200 NVL~174Simulate output on H200 NVLH200 SXM~174Simulate output on H200 SXM192+ GB
Q4_K_M · best fit IQ3_M (182.1 GB)
6 GPUs · 1 Mac class · 2 device families
B200~185Simulate output on B200B300~185Simulate output on B300Instinct MI350X~70Simulate output on Instinct MI350XM3 Ultra256–512 GB~32Simulate output on Mac Studio (M3 Ultra, 256GB)8x H100 SXM5~254Simulate output on NVIDIA DGX H1008x A100 SXM4~231Simulate output on NVIDIA DGX A100 640GB
By VRAM tier · typical GPUs and Macs at Q4_K_M
Consumer GPUs (up to 32 GB) and Macs up to 96 GB: too small for this model (needs ≥ 111.5 GB)8 GB tier →12 GB tier →16 GB tier →24 GB tier →Macs →
Alternatives to Intern S2 397B
Intern-S2 family · 3 models →Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Qwen3.5 397B A17BAlibaba | 112.8 | 242.6 GB | —needs > 24 GB | same size |
| Llama 4 Maverick 17B 128E InstructMeta | 87.1 | 272.7 GB | —needs > 24 GB | same size |
| Llama 3.1 405B InstructMeta | 83.5 | 275.6 GB | —needs > 24 GB | same size |
| MiniMax M3.1 FlashMiniMax | — | 258.6 GB | —needs > 24 GB | same size |
| Ornith 1.5 397Bornith-ai | — | 242.6 GB | —needs > 24 GB | same size |
| Trinity Large ThinkingArcee AI | — | 799.6 GB | —needs > 24 GB | same size |
Specifications
- Publisher
- InternLM
- Family
- Intern-S2
- Parameters
- 403.4B
- Active parameters
- 24.4B per token
- Architecture
- Qwen3_5MoeForConditionalGeneration
- Context Length
- 262,144 tokens
- Vocabulary Size
- 251,392
- Release Date
- 2026-09-13
- License
- Apache 2.0
Frequently Asked Questions
How much VRAM does Intern S2 397B need?
Intern S2 397B requires 242.5 GB of VRAM at Q4_K_M, or 807.3 GB at BF16. Full 262K context adds up to 3.9 GB (246.4 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 403.4B × 4.8 bits ÷ 8 = 242.1 GB
KV Cache + Overhead ≈ 0.4 GB (at 2K context + ~0.3 GB framework)
Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.
KV Cache + Overhead ≈ 4.3 GB (at full 262K context)
VRAM usage by quantization
Can NVIDIA GeForce RTX 5090 run Intern S2 397B?
No — Intern S2 397B requires at least 111.4 GB at IQ2_XXS, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
What's the best quantization for Intern S2 397B?
For Intern S2 397B, Q4_K_M (242.5 GB) offers the best balance of quality and VRAM usage. Q5_K_S (277.8 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 111.4 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run Intern S2 397B on a Mac?
Yes — Mac Pro M2 Ultra (128 GB) and 14 other Macs can run Intern S2 397B, the smallest at IQ2_XS. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 242.5 GB of usable unified memory (RAM minus macOS overhead), from Mac Studio (M3 Ultra, 256GB).
Can I run Intern S2 397B locally?
Yes — Intern S2 397B can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Intern S2 397B?
At Q4_K_M, Intern S2 397B can reach ~66 tok/s on AMD Instinct MI350X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = 1000 ÷ (active GB ÷ (bandwidth GB/s × efficiency) × 1000 + layers × routing ms)
Mixture-of-Experts: just the active experts get read for each token, along with a fixed routing cost per layer.
Example: NVIDIA B300 → 14.7 GB active ÷ (8000 × 0.65) = 2.82 ms, plus 60 layers × 0.055 ms = 3.30 ms, so 1000 ÷ 6.12 ms = ~163 tok/s
Estimated speed at Q4_K_M (242.5 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Intern S2 397B?
At Q4_K_M, the download is about 242.05 GB. The full-precision BF16 version is 806.85 GB. The smallest option (IQ2_XXS) is 110.94 GB.
Which GPUs can run Intern S2 397B?
No single consumer GPU has enough VRAM to run Intern S2 397B at Q4_K_M (242.5 GB). Multi-GPU or professional hardware is required.
Which devices can run Intern S2 397B?
2 devices with unified memory can run Intern S2 397B at Q4_K_M (242.5 GB), including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.