Supported Models & Benchmarks

April 27, 2026 · View on GitHub

This page summarizes all Spatial Intelligence models and benchmarks currently supported by EASI, together with their corresponding model / dataset names for easy reproduction and configuration.

🧠 Supported Models

EASI (backend=VLMEvalKit): Use --model <Model Name> (e.g., --model Qwen2.5-VL-3B-Instruct)

EASI (backend=lmms-eval): Use --model <Base Model> --model_args pretrained=<Link> (e.g., --model qwen2_5_vl --model_args pretrained=Qwen/Qwen2.5-VL-3B-Instruct)

Family	Type	Base Model	Model Name	Link	EASI (backend=VLMEvalKit)	EASI (backend=lmms-eval)
SenseNova-SI 1.0	SI	`internvl3`	`SenseNova-SI-InternVL3-2B`	sensenova/SenseNova-SI-InternVL3-2B	✅	✅
		`internvl3`	`SenseNova-SI-InternVL3-8B`	sensenova/SenseNova-SI-InternVL3-8B	✅	✅
SenseNova-SI 1.1	SI	`qwen2_5_vl`	`SenseNova-SI-1.1-Qwen2.5-VL-3B`	sensenova/SenseNova-SI-1.1-Qwen2.5-VL-3B	✅	✅
		`qwen2_5_vl`	`SenseNova-SI-1.1-Qwen2.5-VL-7B`	sensenova/SenseNova-SI-1.1-Qwen2.5-VL-7B	✅	✅
		`qwen3_vl`	`SenseNova-SI-1.1-Qwen3-VL-8B`	sensenova/SenseNova-SI-1.1-Qwen3-VL-8B	✅	✅
		`internvl3`	`SenseNova-SI-1.1-InternVL3-2B`	sensenova/SenseNova-SI-1.1-InternVL3-2B	✅	✅
		`internvl3`	`SenseNova-SI-1.1-InternVL3-8B`	sensenova/SenseNova-SI-1.1-InternVL3-8B	✅	✅
	UMM	`bagel`	`SenseNova-SI-1.1-BAGEL-7B-MoT`	sensenova/SenseNova-SI-1.1-BAGEL-7B-MoT	✅	✅
SenseNova-SI 1.2	SI	`internvl3`	`SenseNova-SI-1.2-InternVL3-8B`	sensenova/SenseNova-SI-1.2-InternVL3-8B	✅	✅
SenseNova-SI 1.3	SI	`internvl3`	`SenseNova-SI-1.3-InternVL3-8B`	sensenova/SenseNova-SI-1.3-InternVL3-8B	✅	✅
		`qwen3_vl`	`SenseNova-SI-1.3-Qwen3-VL-8B`	sensenova/SenseNova-SI-1.3-Qwen3-VL-8B	✅	✅
SenseNova-SI 1.4	SI	`internvl3`	`SenseNova-SI-1.4-InternVL3-8B`	sensenova/SenseNova-SI-1.4-InternVL3-8B	✅	✅
SenseNova-SI 1.5	SI	`internvl3`	`SenseNova-SI-1.5-InternVL3-8B`	sensenova/SenseNova-SI-1.5-InternVL3-8B	✅	✅
MindCube	SI	`qwen2_5_vl`	`MindCube-3B-RawQA-SFT`	MLL-Lab/MindCube-Qwen2.5VL-RawQA-SFT	✅	✅
		`qwen2_5_vl`	`MindCube-3B-Aug-CGMap-FFR-Out-SFT`	MLL-Lab/MindCube-Qwen2.5VL-Aug-CGMap-FFR-Out-SFT	✅	✅
		`qwen2_5_vl`	`MindCube-3B-Plain-CGMap-FFR-Out-SFT`	MLL-Lab/MindCube-Qwen2.5VL-Plain-CGMap-FFR-Out-SFT	✅	✅
SpatialLadder	SI	`qwen2_5_vl`	`SpatialLadder-3B`	hongxingli/SpatialLadder-3B	✅	✅
SpatialMLLM	SI	`internvl2`	`SpatialMLLM-4B`	Diankun/Spatial-MLLM-subset-sft	✅	✅
SpaceR	SI	`qwen2_5_vl`	`SpaceR-7B`	RUBBISHLIKE/SpaceR	✅	✅
VST	SI	`qwen2_5_vl`	`VST-3B-SFT`	rayruiyang/VST-3B-SFT	✅	✅
		`qwen2_5_vl`	`VST-7B-SFT`	rayruiyang/VST-7B-SFT	✅	✅
Cambrian-S	SI	`cambrians`	`Cambrian-S-0.5B`	nyu-visionx/Cambrian-S-0.5B	✅	✅
		`cambrians`	`Cambrian-S-1.5B`	nyu-visionx/Cambrian-S-1.5B	✅	✅
		`cambrians`	`Cambrian-S-3B`	nyu-visionx/Cambrian-S-3B	✅	✅
		`cambrians`	`Cambrian-S-7B`	nyu-visionx/Cambrian-S-7B	✅	✅
VLM-3R	SI	-	`VLM-3R`	VITA-Group/VLM-3R	✅	❌
BAGEL	UMM	`bagel`	`BAGEL-7B-MoT`	ByteDance-Seed/BAGEL-7B-MoT	✅	✅

📊 Supported Benchmarks

EASI (backend=VLMEvalKit): Use --data <VLMEvalKit Data> (e.g., --data VSI-Bench_32frame)

EASI (backend=lmms-eval): Use --tasks <lmms-eval Task> (e.g., --tasks vsibench)

No	Benchmark	Type	VLMEvalKit Data	lmms-eval Task	Release Date
1	MindCube	image	`MindCubeBench_tiny_raw_qa`, `MindCubeBench_raw_qa`	`mindcube_tiny`, `mindcube_full`	Jun 2025
2	ViewSpatial	image	`ViewSpatialBench`	`viewspatial`	May 2025
3	EmbSpatial-Bench	image	`EmbSpatialBench`	`embspatial`	Jun 2024
4	MMSI-Bench (no circular)	image	`MMSIBench_wo_circular`	`mmsi_bench`	May 2025
5	VSI-Bench	video	`{VSI-Bench}_` `{128frame,64frame,32frame,16frame,2fps,1fps}`	`vsibench`	Dec 2024
6	VSI-Bench-Debiased	video	`{VSI-Bench-Debiased}_` `{128frame,64frame,32frame,16frame,2fps,1fps}`	`vsibench_debiased`	Nov 2025
7	SITE-Bench	image+video	image: `SiteBenchImage` video: `{SiteBenchVideo}_` `{64frame,32frame,1fps}`	`site_bench_image`, `site_bench_video`	May 2025
8	SPAR-Bench	image	`SparBench`, `SparBench_tiny`	`sparbench`, `sparbench_tiny`	Mar 2025
9	STARE-Bench	image	`StareBench`, `StareBench_CoT`	`stare_full`	Jun 2025
10	Spatial-Visualization-Benchmark	image	`SpatialVizBench`, `SpatialVizBench_CoT`	`spatialviz_full`	Jul 2025
11	OmniSpatial	image	`OmniSpatialBench`, `OmniSpatialBench_default`, `OmniSpatialBench_zeroshot_cot`, `OmniSpatialBench_manual_cot`	`omnispatial_test`	Jun 2025
12	ERQA	image	`ERQA`	`erqa`	Apr 2025
13	RefSpatial-Bench	image	`RefSpatial`, `RefSpatial_wo_unseen`	`refspatial`	Jun 2025
14	RoboSpatial-Home	image	`RoboSpatialHome`	-	Nov 2024
15	SPBench	image	`SPBench-MV`, `SPBench-SI`, `SPBench-MV_CoT`, `SPBench-SI_CoT`	-	Oct 2025
16	MMSI-Video-Bench	video	`MMSIVideoBench_` `{300frame,64frame,50frame,32frame,1fps}`	`mmsi_video_u50`	Dec 2025
17	VSI-SUPER-Recall	video	`{VsiSuperRecall}_` `{10mins,30mins,60mins,120mins,240mins}_` `{128frame,64frame,32frame,16frame,2fps,1fps}`	-	Nov 2025
18	VSI-SUPER-Count	video	`{VsiSuperCount}_` `{10mins,30mins,60mins,120mins}_` `{128frame,64frame,32frame,16frame,2fps,1fps}`	-	Nov 2025
19	STI-Bench	video	`{STI-Bench}_` `{64frame,32frame,30frame,1fps}`	-	Apr 2025
20	BLINK	image	`BLINK`, `BLINK_circular`	`blink`	Apr 2024
21	CV-Bench	image	`CV-Bench-2D`, `CV-Bench-3D`	`cv_bench`, `cv_bench_2d`, `cv_bench_3d`	Jun 2024
22	3DSRBench	image	`3DSRBench`	-	Dec 2024
23	LEGO-Puzzles	image	`LEGO`, `LEGO_circular`	-	Mar 2025
24	Spatial457	image	`Spatial457`	-	Feb 2025
25	DSR-Bench	video	`{DSRBench}_` `{64frame,32frame,30frame,1fps}`	-	Dec 2025
26	ERIQ	image	`ERIQ`	-	Dec 2025
27	OSI-Bench	video	`{OSI-Bench}_` `{64frame,32frame,30frame,1fps}`, `{OSI-Bench_visual_first}_` `{64frame,32frame,30frame,1fps}`,	`osi_bench_frames`	Dec 2025