Codesota · Models · InternVL3-78BShanghai AI Lab6 results · 5 benchmarks
Model card

InternVL3-78B.

Shanghai AI Labopen-source78B paramsVision-Language Model
§ 02 · Benchmarks

Every benchmark InternVL3-78B has a recorded score for.

#BenchmarkArea · TaskMetricValueRankDateSource
01MMBenchMultimodal · Visual Question Answeringaccuracy90.1%#3/202025-01-22source ↗
02MME-VideoOCRComputer Vision · General OCR Capabilitiestotal-accuracy67.2%#3/6source ↗
03MMMUMultimodal · Visual Question Answeringaccuracy73.3%#10/302025-01-22unverified
04MMMUMultimodal · Visual Question Answeringaccuracy72.2%#11/30source ↗
05MMMUMultimodal · Image-Text-to-Textaccuracy72.2%#13/36source ↗
06Video-MMEMultimodal · Video Understandingaccuracy72.7%#14/24source ↗
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 03 · Strengths by area

Where InternVL3-78B actually performs.

Computer Vision
1
benchmark
avg rank #3.0
Multimodal
4
benchmarks
avg rank #10.2
§ 04 · Papers

2 papers with results for InternVL3-78B.

  1. 2025-04-14· 3 results

    InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models

  2. 2025-01-22· Multimodal· 2 results

    InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models

§ 05 · Related models

Other Shanghai AI Lab models scored on Codesota.

InternImage-H
2 results · 1 SOTA
Intern-S1-Pro
5 results
InternVL2-76B
76B params · 5 results
InternImage-H
Unknown params · 1 result
InternVL3-76B
1 result
InternVL3.5-241B
1 result
InternImage-XL
0 results
TCP
0 results
§ 06 · Sources & freshness

Where these numbers come from.

pwc-dump
3
results
arxiv
2
results
alphaxiv-leaderboard
1
result
1 of 6 rows marked verified.