ModelRefs / Qwen2.5-VL 72B vs InternVL2.5 78B
Qwen2.5-VL 72B vs InternVL2.5 78B
Qwen2.5-VL 72B vs InternVL2.5 78B: side-by-side benchmarks, pricing, context windows, coding ability and deployment. Pick the right model for your stack.
Overview
InternVL2.5 78B: InternVL2.5 78B is OpenGVLab's multimodal model in the ViT-MLP-LLM configuration, pairing the InternViT-6B-448px-V2.5 vision encoder with Qwen2.5-72B-Instruct as its language model for roughly 78 billion parameters in total. It keeps InternVL 2.0's architecture and advances instead through training and test-time strategy and a larger, filtered data mixture. The project is released under the MIT licence, while the Qwen2.5-72B-Instruct component it embeds carries the Qwen licence.
Qwen2.5-VL 72B: Qwen2.5-VL 72B Instruct is the largest vision-language model in Alibaba's Qwen2.5-VL line, accepting images and video alongside text. Qwen highlights comprehension of videos over an hour long, structured extraction from documents such as invoices, forms and tables, visual grounding returned as bounding boxes or points, and agentic computer and phone use. The published config sets a 32,768-token context, and Qwen advises against YaRN extension here because it degrades temporal and spatial localisation.
Context window: InternVL2.5 78B accepts up to 32,768 tokens, Qwen2.5-VL 72B up to 128,000. Confirm the limit for your specific deployment channel before relying on it.
InternVL2.5 78B vs Qwen2.5-VL 72B at a glance
| Attribute | InternVL2.5 78B | Qwen2.5-VL 72B |
|---|---|---|
| Provider | OpenGVLab | Alibaba |
| Released | 2024-12-06 | 2025-01-26 |
| Context window | 32,768 tokens | 128,000 tokens |
| Input price | Free / self-hosted | Free / self-hosted |
| Output price | Free / self-hosted | Free / self-hosted |
| Licence | MIT | Qwen License |
| Self-hostable | Yes, open weights | Yes, open weights |
| Modalities | text, vision | text, vision |
Benchmarks reported for both
| Benchmark | InternVL2.5 78B | Qwen2.5-VL 72B | Reported |
|---|---|---|---|
| MMMU (accuracy) | 70.1 | 70.2 | 2024-12-06 / 2025-01-26 |
Each row compares the same benchmark, the same reported metric, and the same task set. Scores are as published by their cited sources on the dates shown. Different runs still use different harnesses and prompts, so treat these as directional evidence rather than a head-to-head result.
Where they differ most
- OCR: InternVL2.5 78B 62%, Qwen2.5-VL 72B 87%. Qwen2.5-VL 72B leads on this dimension.
- Multimodal: InternVL2.5 78B 62%, Qwen2.5-VL 72B 75%. Qwen2.5-VL 72B leads on this dimension.
- Cost Efficiency: InternVL2.5 78B 100%, Qwen2.5-VL 72B 100%. Both score the same on this dimension.
Capability scores are ModelRefs' own derived signals, not vendor claims or benchmark results. Validate against your own workload before relying on them.
Continue your research
Use these connected ModelRefs sections to compare alternatives, inspect implementation paths, and review the evidence and governance boundaries relevant to Qwen2.5-VL 72B vs InternVL2.5 78B.
Frequently asked questions
Which is better, Qwen2.5-VL 72B or InternVL2.5 78B?
Qwen2.5-VL 72B and InternVL2.5 78B target different workloads — see the benchmark and pricing tables for a side-by-side answer.
Is Qwen2.5-VL 72B cheaper than InternVL2.5 78B?
Compare $0.00000 vs $0.00000 on this page.