SAVRN
Search Contact SAVRN

SAVRN Model Hub · Comparisons

Florence-2-base vs Qwen3.5-0.8B

Florence-2-base has 232M parameters and Qwen3.5-0.8B has 873M parameters; Florence-2-base is released under MIT License and Qwen3.5-0.8B under Apache License 2.0; at 16-bit, Florence-2-base needs about 0.6 GB (1x MI300X from $1.85 an hour) and Qwen3.5-0.8B about 2.1 GB (1x MI300X from $1.85 an hour).

Published metadata for 2 models, each read from its own repository.
Field Florence-2-base
microsoft/Florence-2-base
Qwen3.5-0.8B
Qwen/Qwen3.5-0.8B
Publisher Microsoft Qwen
Task Image and text to text Image and text to text
Modality Image and text Image and text
Parameters, as reported 232M parameters 873M parameters
Architecture Florence2ForConditionalGeneration Qwen3_5ForConditionalGeneration
Library transformers transformers
Context length 1,024 tokens 262,144 tokens
Repository size 930.3 MB 1.8 GB
Artifact formats safetensors, pytorch safetensors
License mit apache-2.0
Access Open weights, no gate Open weights, no gate
Memory at 16-bit (weights and margin) 0.6 GB 2.1 GB
Cheapest GPUs at 16-bit, per hour 1x MI300X, $1.85 1x MI300X, $1.85
Memory at 4-bit (weights and margin) 0.1 GB 0.5 GB
Cheapest GPUs at 4-bit, per hour 1x MI300X, $1.85 1x MI300X, $1.85
Revision viewed 5ca5edf5bd01 2fc06364715b
Downloads reported by the hub 3M 2.4M
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.

Qwen3.5-0.8B

BenchmarkConditionsResultReported byRevisionDate
Idavidrein/gpqa Task diamondMetric diamondComparison conditions not established 11.9 Model Card
Reported by a third party
Evaluated revision not stated 2026-03-02
LiquidAI/ifstruct-v1.0 Task ifstruct_v1Metric ifstruct_v1Comparison conditions not established 15.5 Liquid AI — IFStruct v1.0 blog (Qwen3.5-0.8B)
Reported by a third party
Evaluated revision not stated 2026-06-30
MMMU/MMMU_Pro Task mmmu_pro_visionMetric mmmu_pro_visionSetup ThinkingComparison conditions not established 31.2 Model Card
Reported by a third party
Evaluated revision not stated 2026-04-28
TIGER-Lab/MMLU-Pro Task mmlu_proMetric mmlu_proComparison conditions not established 29.7 Model Card
Reported by a third party
Evaluated revision not stated 2026-03-02
likaixin/ScreenSpot-Pro Task overallMetric overallComparison conditions not established 46.5 Model Card
Reported by a third party
Evaluated revision not stated 2026-03-18
llamaindex/ExtractBench Task longMetric longSetup Pipeline name: qwen3_5_0_8b_vllm_extract_oneshot_structured_output_fileComparison conditions not established 6.65 ExtractBench
Reported by a third party
Evaluated revision not stated 2026-08-24
llamaindex/ExtractBench Task meanMetric meanSetup Pipeline name: qwen3_5_0_8b_vllm_extract_oneshot_structured_output_fileComparison conditions not established 46.63 ExtractBench
Reported by a third party
Evaluated revision not stated 2026-08-24
llamaindex/ExtractBench Task mediumMetric mediumSetup Pipeline name: qwen3_5_0_8b_vllm_extract_oneshot_structured_output_fileComparison conditions not established 26.98 ExtractBench
Reported by a third party
Evaluated revision not stated 2026-08-24
llamaindex/ExtractBench Task shortMetric shortSetup Pipeline name: qwen3_5_0_8b_vllm_extract_oneshot_structured_output_fileComparison conditions not established 57.44 ExtractBench
Reported by a third party
Evaluated revision not stated 2026-08-24
llamaindex/ParseBench Task chartMetric chartSetup Pipeline name: qwen3_5_0_8b_vllm_layoutComparison conditions not established 0.4 ParseBench
Reported by a third party
Evaluated revision not stated 2026-04-22
llamaindex/ParseBench Task layoutMetric layoutSetup Pipeline name: qwen3_5_0_8b_vllm_layoutComparison conditions not established 15 ParseBench
Reported by a third party
Evaluated revision not stated 2026-04-22
llamaindex/ParseBench Task meanMetric meanSetup Pipeline name: qwen3_5_0_8b_vllm_layoutComparison conditions not established 28.4 ParseBench
Reported by a third party
Evaluated revision not stated 2026-04-22
llamaindex/ParseBench Task tableMetric tableSetup Pipeline name: qwen3_5_0_8b_vllm_layoutComparison conditions not established 1.5 ParseBench
Reported by a third party
Evaluated revision not stated 2026-04-22
llamaindex/ParseBench Task text_contentMetric text_contentSetup Pipeline name: qwen3_5_0_8b_vllm_layoutComparison conditions not established 82 ParseBench
Reported by a third party
Evaluated revision not stated 2026-04-22
llamaindex/ParseBench Task text_formattingMetric text_formattingSetup Pipeline name: qwen3_5_0_8b_vllm_layoutComparison conditions not established 43.1 ParseBench
Reported by a third party
Evaluated revision not stated 2026-04-22

SAVRN's Notes on Florence-2-base

Hand it an image and a short text prompt and it returns text; Microsoft's examples are captioning, object detection and segmentation. At 232M parameters the 16-bit weights are 0.5 GB and the run needs 0.6 GB; 8-bit and 4-bit trim that to 0.3 GB and 0.1 GB. Our Index prices the cheapest qualifying card, a single 192 GB MI300X, at $1.85 an hour on-demand, which this would leave nearly empty, so the question is what else shares it.

MIT is the license: commercial use, modification and redistribution, with the copyright and permission notices kept in the package and nothing else to pass downstream. Access is open. Check the context and the revision. Context is 1,024 tokens: room for a task prompt and a caption back, not long text. The files were last updated 2025-08-04, over a year after the 2024-06-15 release, so pin the revision you tested.

SAVRN's Notes on Qwen3.5-0.8B

Half a gigabyte of memory runs Qwen3.5-0.8B at 4-bit; 16-bit needs 2.1 GB and 8-bit 1.0 GB. It takes images and text in and returns text, with 873 million parameters and a 262,144-token context, and the evaluations the file reports, ScreenSpot-Pro and ExtractBench among them, point at screen and extraction work. The cheapest place to put it is a 192 GB MI300X at $1.85 an hour on demand, where dozens of copies fit, so the hourly rate is the cost that matters, not memory.

The file traces it to Qwen/Qwen3.5-0.8B-Base, and Apache License 2.0 allows commercial use, modification and redistribution, with notices kept and significant changes stated, so a tuned variant of your own is within the terms. Check the long-context path before leaning on the 262,144-token window: the ExtractBench figures in the file fall from 46.63 mean to 26.98 medium to 6.65 long, all third-party reported.

Questions

Which is larger, Florence-2-base or Qwen3.5-0.8B?

Qwen3.5-0.8B (873M parameters) is larger than Florence-2-base (232M parameters), by the parameter counts their publishers report.

Which is cheaper to run, Florence-2-base or Qwen3.5-0.8B?

At 4-bit, Florence-2-base fits on 1x MI300X from $1.85 an hour and Qwen3.5-0.8B on 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.

Can I use Florence-2-base commercially?

Yes. Florence-2-base is released under MIT License. The MIT License is a short permissive license. It permits commercial use, modification and redistribution, provided the copyright notice and permission notice are included.

Can I use Qwen3.5-0.8B commercially?

Yes. Qwen3.5-0.8B 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.

Related Comparisons