SAVRN Model Hub · Comparisons
OpenELM-1_1B-Instruct vs Qwen3-0.6B
OpenELM-1_1B-Instruct has 1.1B parameters and Qwen3-0.6B has 752M parameters; OpenELM-1_1B-Instruct is released under apple-amlr and Qwen3-0.6B under Apache License 2.0; at 16-bit, OpenELM-1_1B-Instruct needs about 2.6 GB (1x MI300X from $1.85 an hour) and Qwen3-0.6B about 1.8 GB (1x MI300X from $1.85 an hour).
| Field | OpenELM-1_1B-Instruct apple/OpenELM-1_1B-Instruct | Qwen3-0.6B Qwen/Qwen3-0.6B |
|---|---|---|
| Publisher | Apple | Qwen |
| Task | Text generation | Text generation |
| Modality | Text | Text |
| Parameters, as reported | 1.1B parameters | 752M parameters |
| Architecture | OpenELMForCausalLM | Qwen3ForCausalLM |
| Library | transformers | transformers |
| Context length | Not stated | 40,960 tokens |
| Repository size | 2.2 GB | 1.5 GB |
| Artifact formats | safetensors | safetensors |
| License | apple-amlr | apache-2.0 |
| Access | Open weights, no gate | Open weights, no gate |
| Memory at 16-bit (weights and margin) | 2.6 GB | 1.8 GB |
| Cheapest GPUs at 16-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Memory at 4-bit (weights and margin) | 0.6 GB | 0.5 GB |
| Cheapest GPUs at 4-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Revision viewed | effd796da2a7 | c1899de289a0 |
| Downloads reported by the hub | 1.4M | 22.5M |
| Last observed | 2026-09-18 | 2026-09-18 |
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.
SAVRN's Notes on OpenELM-1_1B-Instruct
Read the license before the spec sheet on this one. Apple releases it under its own apple-amlr terms, and our facts file carries no summary of what they permit, so legal reads the full text before anyone loads a weight. The hardware question is easy: 1.1 billion parameters need 2.6 GB of memory at 16-bit, 1.3 GB at 8-bit and 0.6 GB at 4-bit, so the cheapest card we price, a 192 GB MI300X at $1.85 an hour, would sit nearly empty serving it alone. It belongs on a shared card with other small models, not on a GPU of its own.
Two gaps to check. No context length is published, so test the input size you need before designing a prompt around it. No host on the SAVRN Index serves it, so there is no rental price to beat; the choice is run it yourself or not at all. Apple ships 270M, 450M, 1.1B and 3B sizes along with the CoreNet training framework, described in arXiv:2404.14619, so the training path is open if you want to adapt it.
SAVRN's Notes on Qwen3-0.6B
The number that stands out here is the context window: 40,960 tokens on a model of 752M parameters. That pairing suits routing, extraction and summarization over long inputs. Memory is not the constraint: 1.8 GB at 16-bit, 0.5 GB at 4-bit. One MI300X with 192 GB, the cheapest setup that runs it at $1.85 per hour on-demand, would hold more than a hundred copies, so it rides beside whatever else you run.
The license is Apache 2.0, so tuning it on your own data and shipping the result is permitted, as long as the notices travel with it and significant changes are stated. It derives from Qwen3-0.6B-Base, so start there if you plan your own post-training, and the Qwen3 technical report, arXiv:2505.09388, sets out the design. No host on our Index quotes a per-token price for it, so plan around the hourly card rate.
Questions
Which is larger, OpenELM-1_1B-Instruct or Qwen3-0.6B?
OpenELM-1_1B-Instruct (1.1B parameters) is larger than Qwen3-0.6B (752M parameters), by the parameter counts their publishers report.
Which is cheaper to run, OpenELM-1_1B-Instruct or Qwen3-0.6B?
At 4-bit, OpenELM-1_1B-Instruct fits on 1x MI300X from $1.85 an hour and Qwen3-0.6B on 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use Qwen3-0.6B commercially?
Yes. Qwen3-0.6B is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.