# Llama-3.2-1B-Instruct vs OpenELM-1_1B-Instruct: GPU Cost Compared
Source: https://savrn.com/models/compare/llama-3-2-1b-instruct-vs-openelm-1-1b-instruct
Markdown alternate of the page above; the site index is https://savrn.com/llms.txt

---

| Field | [Llama-3.2-1B-Instruct](https://savrn.com/models/llama-3-2-1b-instruct) meta-llama/Llama-3.2-1B-Instruct | [OpenELM-1_1B-Instruct](https://savrn.com/models/openelm-1-1b-instruct) apple/OpenELM-1_1B-Instruct |
| --- | --- | --- |
| Publisher | Meta Llama | Apple |
| Task | Text generation | Text generation |
| Modality | Text | Text |
| Parameters, as reported | 1.2B parameters | 1.1B parameters |
| Architecture | LlamaForCausalLM | OpenELMForCausalLM |
| Library | transformers | transformers |
| Context length | Not stated | Not stated |
| Repository size | 5.0 GB | 2.2 GB |
| Artifact formats | safetensors, pytorch | safetensors |
| License | llama3.2 | apple-amlr |
| Access | Access requested at publisher | Open weights, no gate |
| Memory at 16-bit (weights and margin) | 3 GB | 2.6 GB |
| Cheapest GPUs at 16-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Memory at 4-bit (weights and margin) | 0.7 GB | 0.6 GB |
| Cheapest GPUs at 4-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Revision viewed | 9213176726f5 | effd796da2a7 |
| Downloads reported by the hub | 4.6M | 1.1M |
| Last observed | 2026-10-22 | 2026-10-01 |

An evaluation row appears only where at least two of these models report the same benchmark with the same stated configuration, metric, unit and setup. Different evaluators stay named in each cell. Values are shown as reported: no unit conversion, no ranking.

## Other Reported Results

These results are listed for each model on its own, because the conditions needed to compare them are not stated or do not match. Two results that leave a condition blank are not assumed to share it.

### Llama-3.2-1B-Instruct

| Benchmark | Conditions | Result | Reported by | Revision | Date |
| --- | --- | --- | --- | --- | --- |
| Idavidrein/gpqa | Task diamondMetric diamondSetup GPQA DiamondComparison conditions not established | 18.6869 | [EvalEval](https://huggingface.co/datasets/evaleval/EEE_datastore/blob/192329fb7d6b15b7b0936a1a58ae862aa7e8ba24/flat/objects/aa/ad/aaad0db3-7c5b-4834-a1a2-d96624bdd003.json) Reported by a third party | Evaluated revision not stated | 2026-04-16 |

## SAVRN's Notes on Llama-3.2-1B-Instruct

1.2B parameters at 3.0 GB in 16-bit memory, 0.7 GB at 4-bit: cheap to run on the one 192 GB MI300X at $1.85 an hour we list as the cheapest setup — but access is gated, the llama3.2 license is Meta's own rather than Apache or MIT, and the page records no context length, so verify that with your own tests before promising anything.

Meta frames the 1B and 3B instruct pair as multilingual dialogue engines — English, German, French, Italian, Portuguese, Hindi, Spanish and Thai are listed — tuned with SFT and RLHF for agentic retrieval and summarization. The single evaluation on the page, GPQA Diamond at 18.6869, sets the quality ceiling. Weights ship as pth plus safetensors across 13 files totalling 5.0 GB; one attached paper covers SpinQuant quantization, worth reading before you run 4-bit.

SAVRN Research

## SAVRN's Notes on OpenELM-1_1B-Instruct

Read the license before the spec sheet: Apple ships this under its own apple-amlr terms and our facts file summarizes nothing, so legal reads the full text before anyone loads a weight. The spec sheet itself is easy — 1.1B parameters needing 2.6 GB at 16-bit, 1.3 GB at 8-bit, 0.6 GB at 4-bit — so a dedicated MI300X at $1.85 an hour would sit essentially empty. It belongs on a shared card or at the edge.

The gaps are the story: no context length is published, and no SAVRN Index host serves it, so there is no per-token price to benchmark against. What you get is Apple's instruct-tuned entry in a 270M-to-3B family, with the CoreNet training framework open alongside and the paper at arXiv 2404.14619. For research into efficient small models it is a legitimate starting point; for a product, the license review comes first. It draws 1.1M downloads a month.

SAVRN Research

## Questions

### Which is larger, Llama-3.2-1B-Instruct or OpenELM-1_1B-Instruct?

Llama-3.2-1B-Instruct (1.2B parameters) is larger than OpenELM-1_1B-Instruct (1.1B parameters), by the parameter counts their publishers report.

### Which is cheaper to run, Llama-3.2-1B-Instruct or OpenELM-1_1B-Instruct?

At 4-bit, Llama-3.2-1B-Instruct fits on 1x MI300X from $1.85 an hour and OpenELM-1_1B-Instruct on 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.

## Related Comparisons

- [Llama-3.2-1B-Instruct vs Qwen2.5-1.5B-Instruct](https://savrn.com/models/compare/llama-3-2-1b-instruct-vs-qwen2-5-1-5b-instruct)
- [Llama-3.2-1B-Instruct vs TinyLlama-1.1B-Chat-v1.0](https://savrn.com/models/compare/llama-3-2-1b-instruct-vs-tinyllama-1-1b-chat-v1-0)
- [OpenELM-1_1B-Instruct vs Qwen3-0.6B](https://savrn.com/models/compare/openelm-1-1b-instruct-vs-qwen3-0-6b)
- [gemma-3-1b-it vs OpenELM-1_1B-Instruct](https://savrn.com/models/compare/gemma-3-1b-it-vs-openelm-1-1b-instruct)