dev-M26g2 is an open-weight model from Bima Bagaskhoro. It has 3.4B parameters and a 131,072-token context. At 16-bit it needs about 8.2 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index.
Runs On
What it takes to serve dev-M26g2 (3.4B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 6.8 GB | 8.2 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 3.4 GB | 4.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 1.7 GB | 2.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 20, 2026.
dev-M26g2 on every accelerator the SAVRN Index prices, at every precision
Model Card
The publisher has not written a card for this model.
Configuration
- Architecture
- GraniteForCausalLM
- Context length (tokens)
- 131,072
- Layers
- 40
- Hidden size
- 2,560
- Feed-forward size
- 8,192
- Attention heads
- 40
- Key/value heads
- 8
- Vocabulary size
- 100,352
- Model type
- granite
Identity and Version
- Repository
- bimabk/dev-M26g2
- Publisher
- Bima Bagaskhoro
- Task
- Not stated by the source
- Modality
- Other
- Library
- Not stated by the source
- Parameters
- 3.4B parameters
- Languages
- Not stated by the source
- Revision
- 7b564aa2fe052dda98b7364a6a28999b3087d739
- First published
- 2026-09-20
- Last updated
- 2026-09-20
Files and Weights
8 files, 6.8 GB in total. The weights are 1 file totalling 6.8 GB in safetensors.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 6.8 GB | 456efef377c7 |
| config.json | Configuration | 829 B | — |
| generation_config.json | Configuration | 157 B | — |
| chat_template.jinja | Other | 6.1 KB | — |
| loss.txt | Other | 22 B | — |
| .gitattributes | Repository | 1.5 KB | — |
| tokenizer.json | Tokenizer | 7.2 MB | — |
| tokenizer_config.json | Tokenizer | 392 B | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 6.8 GB
Released by Bima Bagaskhoro through its official repository on Hugging Face.
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 6.8 GB |
| 16-bit | 6.8 GB |
| 8-bit | 3.4 GB |
| 4-bit | 1.7 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About dev-M26g2
How much GPU memory does dev-M26g2 need?
About 8.2 GB at 16-bit and 2 GB at 4-bit: the weights (3.4B parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run dev-M26g2 on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
What is dev-M26g2's context length?
131,072 tokens, from the maximum position embeddings in its published configuration.