SAVRN Model Hub · Comparisons
gemma-3-4b-it vs Qwen3-VL-4B-Instruct
Gemma-3-4b-it has 4.3B parameters and Qwen3-VL-4B-Instruct has 4.4B parameters; gemma-3-4b-it is released under Gemma Terms of Use and Qwen3-VL-4B-Instruct under Apache License 2.0; at 16-bit, gemma-3-4b-it needs about 10.3 GB (1x MI300X from $1.85 an hour) and Qwen3-VL-4B-Instruct about 10.7 GB (1x MI300X from $1.85 an hour).
| Field | gemma-3-4b-it google/gemma-3-4b-it | Qwen3-VL-4B-Instruct Qwen/Qwen3-VL-4B-Instruct |
|---|---|---|
| Publisher | Qwen | |
| Task | Image and text to text | Image and text to text |
| Modality | Image and text | Image and text |
| Parameters, as reported | 4.3B parameters | 4.4B parameters |
| Architecture | Gemma3ForConditionalGeneration | Qwen3VLForConditionalGeneration |
| Library | transformers | transformers |
| Context length | Not stated | 262,144 tokens |
| Repository size | 8.6 GB | 8.9 GB |
| Artifact formats | safetensors | safetensors |
| License | gemma | apache-2.0 |
| Access | Access requested at publisher | Open weights, no gate |
| Memory at 16-bit (weights and margin) | 10.3 GB | 10.7 GB |
| Cheapest GPUs at 16-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Memory at 4-bit (weights and margin) | 2.6 GB | 2.7 GB |
| Cheapest GPUs at 4-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Revision viewed | 093f9f388b31 | ebb281ec70b0 |
| Downloads reported by the hub | 1.8M | 3.8M |
| 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.
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.
gemma-3-4b-it
| Benchmark | Conditions | Result | Reported by | Revision | Date |
|---|---|---|---|---|---|
| Idavidrein/gpqa | Task diamondMetric diamondSetup GPQA DiamondComparison conditions not established | 36.8687 | EvalEval Reported by a third party |
Evaluated revision not stated | 2026-04-18 |
| Idavidrein/gpqa | Task mainMetric mainSetup GPQA chain-of-thoughtComparison conditions not established | 16.3677 | EvalEval Reported by a third party |
Evaluated revision not stated | 2026-06-30 |
| sapbot/ask-my-agent-bench-2 | Task fallbackMetric fallbackComparison conditions not established | 32.4 | Not named Reported by a third party |
Evaluated revision not stated | 2026-06-23 |
Qwen3-VL-4B-Instruct
| Benchmark | Conditions | Result | Reported by | Revision | Date |
|---|---|---|---|---|---|
| tiiuae/PBench | Task averageMetric averageSetup Detection-only model combined with SAM2 to convert boxes to segmentation masks.Comparison conditions not established | 49.2 | Community Evals Reported by a third party |
Evaluated revision not stated | 2026-05-11 |
SAVRN's Notes on gemma-3-4b-it
Image and text in, text out, 4.3B parameters. Memory runs 10.3 GB at 16-bit, 5.2 GB at 8-bit and 2.6 GB at 4-bit, so the cheapest Index setup, one MI300X with 192 GB at $1.85 an hour on demand, has most of its card to spare. Plan on it sharing that card with larger tenants.
The license is where to slow down. Gemma Terms of Use allow commercial use and redistribution subject to the Prohibited Use Policy, whose restrictions pass to anyone you distribute to, so treat it as conditional. Access is gated, so the files come from the publisher once you clear it. The page's context field is empty while the publisher describes a 128K window, so confirm it on your stack. It derives from gemma-3-4b-pt. DeepInfra's $0.05 in and $0.10 out per million tokens is the per-token price to beat.
SAVRN's Notes on Qwen3-VL-4B-Instruct
Feed it an image and a question and it answers in text. At 4.4B parameters the memory figures put it on shared hardware: 2.7 GB needed at 4-bit, 5.3 GB at 8-bit, 10.7 GB for the full 16-bit weights. The cheapest Index setup is one MI300X with 192 GB at $1.85 an hour on-demand, a price set by the card rather than the model, so a single copy leaves most of that card idle.
Apache 2.0 permits commercial use, modification and redistribution, so a fine-tuned version can ship inside a product if the license and copyright notices travel with it and significant changes are stated. Confirm the 262,144-token context length, the ceiling on how much image and text goes into one request, and check that the files you receive match the October 11, 2025 release. Four linked technical reports carry the lineage back to the Qwen-VL paper.
Questions
Which is larger, gemma-3-4b-it or Qwen3-VL-4B-Instruct?
Qwen3-VL-4B-Instruct (4.4B parameters) is larger than gemma-3-4b-it (4.3B parameters), by the parameter counts their publishers report.
Which is cheaper to run, gemma-3-4b-it or Qwen3-VL-4B-Instruct?
At 4-bit, gemma-3-4b-it fits on 1x MI300X from $1.85 an hour and Qwen3-VL-4B-Instruct on 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use gemma-3-4b-it commercially?
Yes, with conditions. gemma-3-4b-it is released under Gemma Terms of Use. Gemma models are released under Google's Gemma Terms of Use, which permit commercial use and redistribution subject to the Gemma Prohibited Use Policy, whose restrictions must be passed on to anyone the model is distributed to.
Can I use Qwen3-VL-4B-Instruct commercially?
Yes. Qwen3-VL-4B-Instruct 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.