SAVRN Model Hub · Comparisons
Qwen2.5-VL-32B-Instruct vs Qwen3.6-35B-A3B
Qwen2.5-VL-32B-Instruct has 33.5B parameters and Qwen3.6-35B-A3B has 36B parameters; both are released under Apache License 2.0; at 16-bit, Qwen2.5-VL-32B-Instruct needs about 80.3 GB (1x MI300X from $1.85 an hour) and Qwen3.6-35B-A3B about 86.3 GB (1x MI300X from $1.85 an hour).
| Field | Qwen2.5-VL-32B-Instruct Qwen/Qwen2.5-VL-32B-Instruct | Qwen3.6-35B-A3B Qwen/Qwen3.6-35B-A3B |
|---|---|---|
| Publisher | Qwen | Qwen |
| Task | Image and text to text | Image and text to text |
| Modality | Image and text | Image and text |
| Parameters, as reported | 33.5B parameters | 36B parameters |
| Architecture | Qwen2_5_VLForConditionalGeneration | Qwen3_5MoeForConditionalGeneration |
| Library | transformers | transformers |
| Context length | 128,000 tokens | 262,144 tokens |
| Repository size | 68.3 GB | 71.9 GB |
| Artifact formats | safetensors | safetensors |
| License | apache-2.0 | apache-2.0 |
| Access | Open weights, no gate | Open weights, no gate |
| Memory at 16-bit (weights and margin) | 80.3 GB | 86.3 GB |
| Cheapest GPUs at 16-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Memory at 4-bit (weights and margin) | 20.1 GB | 21.6 GB |
| Cheapest GPUs at 4-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Revision viewed | 7cfb30d71a1f | 995ad96eacd9 |
| Downloads reported by the hub | 1.1M | 3.7M |
| Last observed | 2026-09-22 | 2026-10-16 |
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.
Qwen3.6-35B-A3B
| Benchmark | Conditions | Result | Reported by | Revision | Date |
|---|---|---|---|---|---|
| Idavidrein/gpqa | Task diamondMetric diamondComparison conditions not established | 86 | Model Card Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| MMMU/MMMU_Pro | Task mmmu_pro_visionMetric mmmu_pro_visionComparison conditions not established | 75.3 | Model Card Reported by a third party |
Evaluated revision not stated | 2026-05-15 |
| MathArena/aime_2026 | Task MathArena/aime_2026Metric MathArena/aime_2026Comparison conditions not established | 92.7 | Model Card Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| MathArena/hmmt_feb_2026 | Task MathArena/hmmt_feb_2026Metric MathArena/hmmt_feb_2026Comparison conditions not established | 83.6 | Model Card Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| SWE-bench/SWE-bench_Multilingual | Task swe_bench_multilingual_%_resolvedMetric swe_bench_multilingual_%_resolvedComparison conditions not established | 67.2 | Model Card Reported by a third party |
Evaluated revision not stated | 2026-08-10 |
| SWE-bench/SWE-bench_Verified | Task swe_bench_%_resolvedMetric swe_bench_%_resolvedComparison conditions not established | 73.4 | Model Card Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| ScaleAI/SWE-bench_Pro | Task SWE_Bench_ProMetric SWE_Bench_ProComparison conditions not established | 49.5 | Model Card Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| TIGER-Lab/MMLU-Pro | Task mmlu_proMetric mmlu_proComparison conditions not established | 85.2 | Model Card Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| cais/hle | Task hleMetric hleComparison conditions not established | 21.4 | Model Card Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| harborframework/terminal-bench-2.0 | Task terminalbench_2Metric terminalbench_2Setup Harbor/Terminus-2 harness; 3h timeout, 32 CPU/48 GB RAM; temp=1.0, top_p=0.95, top_k=20, max_tokens=80K, 256K ctx; avg of 5 runs.Comparison conditions not established | 51.5 | Model Card Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| llamaindex/ExtractBench | Task longMetric longSetup Pipeline name: qwen3_6_35b_a3b_fp8_vllm_extract_oneshot_structured_output_file; served checkpoint Qwen/Qwen3.6-35B-A3B-FP8 on vLLM, one-shot json_object structured outputComparison conditions not established | 36.9 | ExtractBench Reported by a third party |
Evaluated revision not stated | 2026-08-26 |
| llamaindex/ExtractBench | Task meanMetric meanSetup Pipeline name: qwen3_6_35b_a3b_fp8_vllm_extract_oneshot_structured_output_file; served checkpoint Qwen/Qwen3.6-35B-A3B-FP8 on vLLM, one-shot json_object structured outputComparison conditions not established | 88.11 | ExtractBench Reported by a third party |
Evaluated revision not stated | 2026-08-26 |
| llamaindex/ExtractBench | Task mediumMetric mediumSetup Pipeline name: qwen3_6_35b_a3b_fp8_vllm_extract_oneshot_structured_output_file; served checkpoint Qwen/Qwen3.6-35B-A3B-FP8 on vLLM, one-shot json_object structured outputComparison conditions not established | 86.37 | ExtractBench Reported by a third party |
Evaluated revision not stated | 2026-08-26 |
| llamaindex/ExtractBench | Task shortMetric shortSetup Pipeline name: qwen3_6_35b_a3b_fp8_vllm_extract_oneshot_structured_output_file; served checkpoint Qwen/Qwen3.6-35B-A3B-FP8 on vLLM, one-shot json_object structured outputComparison conditions not established | 92.86 | ExtractBench Reported by a third party |
Evaluated revision not stated | 2026-08-26 |
| llamaindex/ParseBench | Task chartMetric chartSetup Pipeline name: qwen3_6_35b_a3b_fp8_vllm_parse_layoutComparison conditions not established | 5.1 | ParseBench Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| llamaindex/ParseBench | Task layoutMetric layoutSetup Pipeline name: qwen3_6_35b_a3b_fp8_vllm_parse_layoutComparison conditions not established | 47.4 | ParseBench Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| llamaindex/ParseBench | Task meanMetric meanSetup Pipeline name: qwen3_6_35b_a3b_fp8_vllm_parse_layoutComparison conditions not established | 44.1 | ParseBench Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| llamaindex/ParseBench | Task tableMetric tableSetup Pipeline name: qwen3_6_35b_a3b_fp8_vllm_parse_layoutComparison conditions not established | 19.1 | ParseBench Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| llamaindex/ParseBench | Task text_contentMetric text_contentSetup Pipeline name: qwen3_6_35b_a3b_fp8_vllm_parse_layoutComparison conditions not established | 90.7 | ParseBench Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
| llamaindex/ParseBench | Task text_formattingMetric text_formattingSetup Pipeline name: qwen3_6_35b_a3b_fp8_vllm_parse_layoutComparison conditions not established | 58.3 | ParseBench Reported by a third party |
Evaluated revision not stated | 2026-04-16 |
SAVRN's Notes on Qwen2.5-VL-32B-Instruct
Run this at 16-bit and plan for 80.3 GB: 66.9 GB of weights plus runtime, fitting one MI300X with 192 GB at $1.85 an hour, the cheapest setup we list. At 8-bit the need falls to 40.1 GB, at 4-bit to 20.1 GB, where an H100 at $1.99 also fits. The 33.5B parameters take image and text in and return text. Qwen cites a 128,000-token context while the configuration lists a 32,768-token sliding window, so test your longest inputs before sizing the KV cache.
Apache 2.0 permits commercial use with notices kept. The download line says the release comes through ModelScope, and the pull is 68.3 GB across 18 safetensors shards — plan the mirror step. The page records an AWQ quantization built from this checkpoint if you want a pre-cut 4-bit, and cites the YaRN context-extension paper alongside technical report arXiv 2502.13923. At 1.1M downloads a month it is one of the most-deployed open vision-language models in this class.
SAVRN's Notes on Qwen3.6-35B-A3B
Memory for 36B parameters, compute for the 3B that fire per token — that is the MoE trade in the A3B name, and it is why one MI300X with 192 GB at $1.85 an hour holds the full 16-bit working set of 86.3 GB with room for the 262,144-token window. At 4-bit the footprint is 21.6 GB, leaving most of the card for KV cache. The listing is image-and-text to text, the publisher's summary leads with coding, and third-party-reported evals are strong for the size: 86 on GPQA-diamond, 92.7 on AIME 2026, 73.4 on SWE-bench Verified.
Renting is priced against owning: DeepInfra lists $0.10 in and $0.95 out per million tokens, Scaleway $0.29 and $1.71 — run your monthly volume against the $1.85 card hour before committing. Apache 2.0 allows commercial use and modification with the notices kept. The 71.9 GB of safetensors span 40 files, released April 2026 and updated days later, so check the revision you pull; the MoE sibling tradeoff on this list is the dense Qwen3.6-27B, which posts higher evals at a similar card price.
Questions
Which is larger, Qwen2.5-VL-32B-Instruct or Qwen3.6-35B-A3B?
Qwen3.6-35B-A3B (36B parameters) is larger than Qwen2.5-VL-32B-Instruct (33.5B parameters), by the parameter counts their publishers report.
Which is cheaper to run, Qwen2.5-VL-32B-Instruct or Qwen3.6-35B-A3B?
At 4-bit, Qwen2.5-VL-32B-Instruct fits on 1x MI300X from $1.85 an hour and Qwen3.6-35B-A3B on 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use Qwen2.5-VL-32B-Instruct commercially?
Yes. Qwen2.5-VL-32B-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.
Can I use Qwen3.6-35B-A3B commercially?
Yes. Qwen3.6-35B-A3B 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.