Granite Switch 4.1 8B
- Open weights
Prices
0 offerings · prices as listedNo priced offering is listed for this model yet.
Details
- Licence
- apache-2.0
- Family
- granite-switch
Not stated on IBM’s pages: max input, input modalities, output modalities, reasoning, reasoning efforts, reasoning can be off, max thinking budget, tool use, structured output, built-in tools, knowledge cutoff, aliases.
Run locally
Weights: ibm-granite/granite-switch-4.1-8b- Parameters 8.8B
- Dense
- Native precision BF16
- Layers 42
- KV cache per token 168 KB
- Config positions 128K (config.json max_position_embeddings)
- Needs
- ≈ 18.5 GB (weights 16.3 (maker’s file) + KV cache 1.3 + overhead 0.8)
- Usable
- 21.6 GB (90% of its memory)
- Generation speed
- 27–40 tok/s
- Longest context
- 27K (longer doesn’t fit)
- First token
- about 0.28–0.43 s for a 2K prompt
| Precision | Weights | Needs at 8K | On your hardware | Longest context |
|---|---|---|---|---|
| BF16 (native)maker’s file | 16.3 GB | ≈ 18.5 | ●Runs well 27–40 tok/s | 27K |
| Q8_0maker’s file | 9.4 GB | ≈ 11.2 | ●Runs well 44–66 tok/s | 71K |
| FP8 | 8.2 GB | ≈ 10.0 | ●Runs well 49–74 tok/s | 78K |
| Q6_Kmaker’s file | 7.5 GB | ≈ 9.3 | ●Runs well 53–80 tok/s | 82K |
| Q5_K_Mmaker’s file | 6.7 GB | ≈ 8.5 | ●Runs well 58–88 tok/s | 87K |
| Q5_0maker’s file | 6.5 GB | ≈ 8.3 | ●Runs well 60–90 tok/s | 89K |
| Q4_K_Mmaker’s file | 6.0 GB | ≈ 7.8 | ●Runs well 64–97 tok/s | 92K |
| Q4_0maker’s file | 5.6 GB | ≈ 7.4 | ●Runs well 68–102 tok/s | 94K |
| IQ4_XS | 4.3 GB | ≈ 6.2 | ●Runs well 83–125 tok/s | 102K |
| Q3_K_Mmaker’s filequality drops | 5.1 GB | ≈ 6.9 | ●Runs well 73–110 tok/s | 97K |
| IQ3_XXSquality drops | 3.1 GB | ≈ 4.9 | ●Runs well 106–159 tok/s | 109K |
| Q2_Kmaker’s filequality drops | 4.3 GB | ≈ 6.2 | ●Runs well 83–125 tok/s | 102K |
| IQ2_XXSquality drops | 2.1 GB | ≈ 3.9 | ●Runs well 137–206 tok/s | 116K |
Click a row to choose a precision. “Maker’s file” sizes are the maker’s own files; others are parameters × bits per weight.
Compared with other hardware
Each at the highest precision it runs. Click one for details.
Smallest that runs it (Q4_K_M, 8K): GPU GeForce RTX 3080; unified memory Apple M5 · 16 GB.
Run commands
vllm serve ibm-granite/granite-switch-4.1-8b --max-model-len 8192
Memory to fine-tune
Batch 1, 2,048-token sequences, gradient checkpointing; LoRA and QLoRA train rank-16 adapters on the attention projections.
Estimates; facts from the maker’s Hugging Face repositories, as of 2026-11-06. How it’s estimated
Cited scores
From third-party leaderboards, not measured by usThis model isn’t on any leaderboard we cite yet.
News
All news →- Oct 14, 2026Release IBM releases Granite Switch 4.1 8B Granite Switch 4.1 8B is now available.
Changes
All changes →- New language model
Sources
- ibm-granite/granite-switch-4.1-8b config.json (opens in a new tab) huggingface.co · checked Nov 2, 2026
- ibm-granite/granite-switch-4.1-8b model card (opens in a new tab) huggingface.co · checked Nov 2, 2026
- ibm-granite/granite-switch-4.1-8b model card (opens in a new tab) huggingface.co · checked Nov 2, 2026
- ibm-granite/granite-switch-4.1-8b files (opens in a new tab) huggingface.co · checked Nov 2, 2026
- ibm-granite/granite-switch-4.1-8b-GGUF files (opens in a new tab) huggingface.co · checked Nov 2, 2026
- ibm-granite/granite-switch-4.1-8b · Hugging Face (opens in a new tab) huggingface.co · checked Nov 3, 2026
Some values on this page are still waiting to be checked against official pages: a filled dot is verified, a hollow one is not yet (with or without a source on file). How we verify data
IBM · United States · ibm.com
Related models
Snapshots, the same family and other models from this provider