// InternLM / Intern-S · Text LLM
internlm3-8b-instruct: VRAM & local deployment
8.8B parameters. About 6.8 GB at 4-bit and 20.3 GB at BF16 with one 8K-token request.
released 2025-01-13licence: Commercial OKmemory figures are estimates
01Specs
| Publisher | Shanghai AI Lab · internlm |
|---|---|
| Family / series | InternLM / Intern-S · internlm3 |
| Type | Text LLM (text-generation) |
| Parameters | 8.8B Source: counted from official safetensors weights |
| Context length | 32K tokens (32,768)max_position_embeddings in config.json |
| Architecture | 48 layers · GQA |
| Released weights | BF16 · 17.6 GBdownload |
| Licence | Apache-2.0 · Commercial OKPermissive licence (Apache-2.0 / MIT / CC-BY / OpenMDW): commercial use allowed. |
| Released | 2025-01-13 (Hugging Face repo created) |
02How much VRAM it needs (estimate)
Total = weights + 10% + 0.5GB overhead + KV cache for one request (FP16 KV). More concurrent users multiply the KV column.
| Precision | Weights | Total @ 8K | Total @ 32K | Smallest GPU | Mac |
|---|---|---|---|---|---|
| BF16 / FP16 (original) | 17.6 GB | 20.3 GB | 21.5 GB | RTX 4090 24GB | Mac 36GB |
| FP8 / INT8 / Q8_0 | 9.3 GB | 11.2 GB | 12.4 GB | 16GB GPU | Mac 24GB |
| 4-bit (Q4_K_M / AWQ / GPTQ-Int4) | 5.3 GB | 6.8 GB | 8.0 GB | 8GB GPU | Mac 16GB |
KV cache per request: 8K 0.4 GB · 32K 1.6 GB
Bits per weight: BF16 16, FP8 8.5, 4-bit 4.85. “Fits” = total ≤ 90% of memory; Mac GPU can use ~2/3 of unified memory (≤36GB) or ~3/4 (larger). For MoE, all experts must be in memory, so size by total parameters. Method →
03How to run it
Only runtimes the publisher or the runtime’s official library names for this model.
- vLLM — mentioned in the official model card
- Ollama — mentioned in the official model card
- Transformers — mentioned in the official model card
- LMDeploy — mentioned in the official model card
- llama.cpp — official GGUF weights
04Official repositories & quantised builds
4 official repo(s) from internlm.
- internlm3-8b-instruct Instruct
- internlm3-8b-instruct-gguf Instruct/GGUF
- internlm3-8b-instruct-awq Instruct/AWQ
- internlm3-8b-instruct-gptq-int4 Instruct/GPTQ/INT4
05Other InternLM / Intern-S models
All 12 →| Model | Params | 4-bit total | Released |
|---|---|---|---|
| Intern-S2-397B | 403.4B | 270 GB | 2026-09-13 |
| Atria-Dawn-Preview | 753.3B | 504 GB | 2026-09-11 |
| Intern-S2-Mobius | 36B | 24.7 GB | 2026-07-29 |
| Intern-S2-Preview | 36.1B | 24.7 GB | 2026-05-15 |
| Intern-S1-Pro | 1T / 22B | 669 GB | 2026-02-02 |
| Intern-S1-mini | 8.54B | 7.4 GB | 2025-08-18 |
| Intern-S1 | 240.7B | 163 GB | 2025-07-24 |
| internlm2_5-1_8b-chat | 1.89B | 2.6 GB | 2024-07-30 |
06Sources
Data checked 2025-02-11. Parameters, context and licence are copied from the official repository; memory figures are our estimates.