SOTA Spatial Reasoning on VSI-Bench and PapersWithCode

Human

2026.02

95.8

47

60.4

94.7

94.3

100

79.2

-

95.8

45.9

-

Human

2026.03

95.8

-

94.7

-

100

79.2

-

95.8

45.9

94.3

60.4

-

47

-

Human

2026.04

95.8

47

-

94.7

-

79.2

-

95.8

45.9

94.3

60.4

-

100

-

Human

2026.06

95.8

-

94.7

-

100

79.2

-

95.8

45.9

94.3

60.4

-

47

-

Human

2026.06

95.8

47

-

94.7

-

100

79.2

-

95.8

45.9

94.3

60.4

-

SSR-3D

2026.02

93.4

67.1

76

71.3

70.5

85

73.9

-

48.5

79.5

-

Cambrian-P

2026.05

89.5

-

74.8

-

85

73.7

-

52.6

76.9

74.9

76

-

60.1

-

GS-Reasoner (pred dep.)

2026.02

88.9

61.9

70

65.4

69.1

52.3

64.7

-

44.3

65.7

-

SpaceMind-8B

2026.06

88.4

-

67.2

-

70.6

69.6

-

44.3

74.2

73.3

77.3

-

61.4

-

SpaceMind

2026.06

88.4

61.4

-

67.2

-

70.6

69.6

-

44.3

74.2

73.3

77.3

-

SSR-2D

2026.02

87.9

63.4

76.5

69.2

71.8

81.9

71.9

-

48.5

76.5

-

VLM²-7B

2026.05

87.8

59.6

-

69

-

68.3

68.8

-

52.6

69.9

72.5

70.8

-

Spatial Code-4B

2026.06

87.1

60.7

-

62

-

59

60

-

32.5

33.1

92

50.8

-

Q-GeoMem

2026.05

85.5

54.5

-

71.4

-

73.4

70.1

-

46.9

76.3

76.7

-

OneCanvas

2026.06

84.8

59.5

-

71

-

65.7

70.1

-

59.8

76.5

68.4

75.4

-

SpatialStack-5B

2026.03

84.1

55.6

-

67.3

-

83.5

67.5

-

41.2

68.2

71

69.1

-

SpatialStack-5B

2026.06

84.1

-

67.3

-

83.5

67.5

-

41.2

68.2

71

69.1

-

55.6

-

Cambrian-S-7B†

2026.05

82

-

71.5

-

84.3

69.2

-

38.7

74.7

73.6

75.2

-

53.7

-

SenseNova-SI

2026.06

80.8

53.5

-

69.6

-

76.4

68.8

-

48.5

72.8

72

76.8

-

SenseNova-SI InternVL3-8B

2026.02

80.7

72

72.7

76.3

76.7

48.4

68.7

-

69.5

53.5

-

GeoVR-2B

2026.06

80.7

-

71.3

-

86.7

69.1

-

45.9

72.3

67.7

73.9

-

54.5

-

VLM-3R-7B

2026.02

80.5

49.4

69.2

65.4

70.2

40.1

60.9

-

45.4

67.1

-

VLM-3R (7B)

2025.05

80.5

-

65.4

-

40.1

60.9

-

45.4

67.1

70.2

69.2

-

49.4

-

VLM-3R-7B

2026.04

80.5

49.4

-

65.4

-

60.9

-

45.4

67.1

70.2

69.2

-

40.1

-

VLM-3R 7B

2026.05

80.5

-

65.4

-

40.1

60.9

-

45.4

67.1

70.2

69.2

-

49.4

-

VLM-3R-7B

2026.05

80.5

49.4

-

65.4

-

40.1

60.9

-

45.4

67.1

70.2

69.2

-

VLM-3R

2026.05

80.5

-

45.4

-

70.2

-

VLM-3R-7B

2026.06

80.5

-

65.4

-

40.1

60.9

-

45.4

67.1

70.2

69.2

-

49.4

-

VLM-3R-7B

2026.06

80.5

49.4

-

65.4

-

40.1

60.9

-

45.4

67.1

70.2

69.2

-

VLM-3R

2026.06

80.5

49.4

-

65.4

-

40.1

60.9

-

45.4

67.1

70.2

69.2

-

Cambrian-S-7B

2026.02

76.2

50.5

74.9

71.1

73.2

80.1

67.5

-

41.8

72.2

-

Cambrian-S 7B

2026.05

76.2

-

71.1

-

80.1

67.5

-

41.8

72.2

73.2

74.9

-

50.5

-

Cambrian-S-7B

2026.06

76.2

-

71.1

-

80.1

67.5

-

41.8

72.2

73.2

74.9

-

50.5

-

Cambrian-S

2026.06

76.2

50.5

-

71.1

-

80.1

67.5

-

41.8

72.2

73.2

74.9

-

Spatial-TTT-2B

2026.03

73

-

61.8

-

77

64.4

-

47.4

65.9

70.8

71.7

-

47.8

-

SpaR3D-MoE

2026.07

70.1

48

-

58.7

-

73.5

63.5

-

44

69.6

71.3

72.8

-

SpatialImaginer

2026.04

69.9

46.8

-

67.4

-

63.7

-

46.4

64.6

70.6

72.7

-

71.2

-

Cambrian-S-7B

2026.04

69.6

45.8

-

66.8

-

62.9

-

39.2

67.6

68.2

72.5

-

73.8

-

Cambrian-S-7B

2026.06

69.6

45.8

-

66.8

-

73.8

62.9

-

39.2

67.6

68.2

72.5

-

LLaVA-Next-Video-7B (Finetuned)

2025.05

68.9

-

64.9

-

38.5

57.7

-

40.7

63.7

70.6

70.8

-

43.6

-

SpatialStack-4B

2026.03

68.4

45.4

-

57.9

-

79.6

60.9

-

40.2

63.2

69.2

63

-

SpatialStack-4B

2026.06

68.4

-

57.9

-

79.6

60.9

-

40.2

63.2

69.2

63

-

45.4

-

Skill-3D

2026.06

67

-

62.7

-

78.1

-

71.4

77.6

66.2

74.9

-

61.5

-

LLaVA-One-Vision-7B (Finetuned)

2025.05

65.9

-

59.7

-

39.4

55.8

-

41.7

59.5

68.3

66.5

-

45.3

-

Skill-3D

2026.06

64.2

-

67.6

-

76.4

-

69

72.8

62.4

73.1

-

58

-

Ouro-Spatial-4B

2026.06

63.7

47.5

-

63.8

-

71.4

62.7

-

33

73.5

71.5

77.1

-

Skill-3D

2026.06

63

-

60.4

-

73.4

-

67.5

71.8

60.9

71.2

-

56.1

-

Ouro-Spatial-8B

2026.06

62.6

48.7

-

64.2

-

70.7

63.3

-

35.6

75.7

71.7

77.3

-

GR3D

2026.03

62.2

-

57.6

-

68.1

58.7

-

61.2

60.7

53.4

63.5

-

42.7

-

Cambrian-S-3B

2026.03

61.9

-

64.8

-

78.8

57.3

-

27.3

46.3

70.7

68

-

40.6

-

Cambrian-S-3B

2026.03

61.9

40.6

-

64.8

-

78.8

57.3

-

27.3

46.3

70.7

68

-

Cambrian-S-3B

2026.06

61.9

-

64.8

-

78.8

57.3

-

27.3

46.3

70.7

68

-

40.6

-

Cambrian-S-3B

2026.06

61.9

40.6

-

64.8

-

78.8

57.3

-

27.3

46.3

70.7

68

-

Think3D

2026.06

61.8

-

58.3

-

73.8

-

66.5

69.9

61.1

70.8

-

55

-

Think3D

2026.06

60.8

-

56.5

-

69.3

-

64.2

66.3

56.8

68

-

52.3

-

Skill-3D-8B

2026.06

60.1

48.6

-

52

-

66.8

-

58.4

67.9

56.5

59.8

-

Think3D

2026.06

59.5

-

64.8

-

72.2

-

65.8

66.4

58.2

69.5

-

53.1

-

Skill-3D

2026.06

59.2

-

53.4

-

35.4

-

62.7

69.5

56.8

58.1

-

42.6

-

Gemini-3-pro

2026.03

57.5

-

56.6

-

60

56

-

61.9

41.8

49

71.5

-

42.8

-

Qwen3.5-4B

2026.03

57.5

36.5

-

60.3

-

62.3

53.6

-

34

53.8

56.5

67.5

-

Qwen3.5-4B

2026.06

57.5

-

60.3

-

62.3

53.6

-

34

53.8

56.5

67.5

-

36.5

-

Gemini-3-Pro

2026.06

57.5

42.8

-

56.6

-

68

56

-

61.9

41.8

49

71.5

-

ReMoT -CoT-4B

2026.02

57.4

42.5

-

52.1

-

59.7

58.8

-

45.6

56.3

78.2

76.8

1

-

VST-7B-SFT

2026.03

55.8

-

59.7

-

65.2

60.6

-

44.9

68.3

72

74.3

-

44.4

-

VST-7B

2026.05

55.8

44.4

-

59.7

-

65.2

60.6

-

44.9

68.3

72

74.3

-

VST-7B-SFT

2026.06

55.8

-

59.7

-

65.2

60.6

-

44.9

68.3

72

74.3

-

44.4

-

VST-7B-SFT

2026.06

55.8

44.4

-

59.7

-

65.2

60.6

-

44.9

68.3

72

74.3

-

Gemini-3-Pro

2026.04

55.7

37.8

-

59.9

-

52.5

-

45.9

44.1

38

72.7

-

66

-

VST-7B

2026.02

55.6

43.8

75.5

60

71.6

69.2

61.2

-

44.3

69.2

-

VST 7B

2026.05

55.6

-

60

-

69.2

61.2

-

44.3

69.2

71.6

75.5

-

43.8

-

VST

2026.06

55.6

43.8

-

60

-

69.2

61.2

-

44.3

69.2

71.6

75.5

-

w/ Tools

2026.06

53.3

-

51.7

-

74.6

-

52.5

58.2

58

69.3

-

47.8

-

Think3D-8B

2026.06

52.9

38.5

-

41.6

-

60.8

-

45.4

58.1

48.3

51.2

-

Think3D

2026.06

52.6

-

47.9

-

29.2

-

55.9

61.4

50.4

50.5

-

32.7

-

Qwen3-VL-8B

2026.06

51.5

47.1

-

58.3

-

64.2

56.5

-

31.4

57.2

66

75.9

-

w/ Tools

2026.06

51.2

-

62.6

-

72.7

-

49.5

46.3

48.4

66

-

41.5

-

Qwen3-VL-8B-Instruct

2026.04

51

47

-

58

-

57.9

-

35

61.9

67.6

76.3

-

66.3

-

Qwen3-VL-8B-Instruct

2026.02

50.9

47

76.3

58

53.3

66.3

57.9

-

35

61.9

-

Qwen3-VL-8B-Instruct

2026.03

50.9

-

58

-

66.3

57.9

-

35

61.9

67.5

76.3

-

47

-

Qwen3-VL-8B-Instruct

2026.06

50.9

-

58

-

66.3

57.9

-

35

61.9

67.5

76.3

-

47

-

w/o Tools

2026.06

50.5

-

48.4

-

73.2

-

49.1

55.7

55.8

67.3

-

43.6

-

Skill-3D-4B

2026.06

50.4

36.8

-

43.5

-

56.7

-

48.8

57.4

48.6

50.2

-

Robust-TO (Qwen3-VL-7B-Instruct)

2026.06

48.8

-

55.5

-

56.4

-

39.8

-

77.5

GPT-5

2026.04

48.7

34.5

-

63.7

-

55

-

50.3

47.5

53.3

73.3

-

68.9

-

VST-7B-SFT

2026.04

48.7

37.3

-

55.2

-

55.5

-

41.8

62.2

68.8

74.5

-

55.5

-

GPT-5-2025-08-07

2026.02

48.6

34.4

73.3

63.7

53.5

68.9

55

-

50.2

47.5

-

GPT-5

2026.03

48.6

-

63.7

-

68.9

55

-

50.2

47.5

53.3

73.3

-

34.4

-

GPT-5

2026.06

48.6

-

63.7

-

68.9

55

-

50.2

47.5

53.3

73.3

-

34.4

-

GPT-5

2026.06

48.6

34.4

-

47.8

-

68.9

55

-

50.2

48.3

53.3

73.3

-

w/ Tools

2026.06

48.4

27.8

-

44.7

-

52.6

-

43.2

46

44.7

48.1

-

Spatial-MLLM-4B

2026.04

48.2

38.1

-

40.4

-

46.3

-

33

35.5

66.7

63.6

-

44.3

-

Gemini-1.5-Pro

2026.03

48.1

28.8

-

46

-

68

48.8

-

42

49.4

49.6

58.6

-

Random Choice

2026.03

47.9

-

25.1

-

25.2

34

-

28.4

33.1

62.1

29.9

-

32

-

Chance Level (Frequency)

2026.03

47.9

32

-

25.1

-

25.2

34

-

28.4

33.1

62.1

29.9

-

Chance Level (Frequency)

2025.05

47.9

-

25.1

-

25.2

34

-

28.4

33.1

62.1

29.9

-

32

-

Random Choice(Frequency)

2026.04

47.9

32

-

25.1

-

34

-

28.4

33.1

62.1

29.9

-

25.2

-

Chance Level (Frequency)

2026.05

47.9

-

25.1

-

25.2

34

-

28.4

33.1

62.1

29.9

-

32

-

Random

2026.06

47.9

32

-

25.1

-

25.2

34

-

28.4

33.1

62.1

29.9

-

Gemini-2.5 Pro

2026.03

47.8

34.9

-

61.1

-

71.3

51.5

-

45.9

42.8

43.8

64.3

-

Gemini-2.5 Pro

2025.05

47.8

-

61.1

-

71.3

51.5

-

45.9

42.8

43.8

64.3

-

34.9

-

Method
Human 2026.02	95.8	47	60.4	94.7	94.3	100	79.2	-	95.8	45.9	-	-	-	-	-	-	-	-	-	-	-
Human 2026.03	95.8	-	-	94.7	-	100	79.2	-	95.8	45.9	94.3	60.4	-	47	-	-	-	-	-	-	-
Human 2026.04	95.8	47	-	94.7	-	-	79.2	-	95.8	45.9	94.3	60.4	-	-	-	100	-	-	-	-	-
Human 2026.06	95.8	-	-	94.7	-	100	79.2	-	95.8	45.9	94.3	60.4	-	47	-	-	-	-	-	-	-
Human 2026.06	95.8	47	-	94.7	-	100	79.2	-	95.8	45.9	94.3	60.4	-	-	-	-	-	-	-	-	-
SSR-3D 2026.02	93.4	67.1	76	71.3	70.5	85	73.9	-	48.5	79.5	-	-	-	-	-	-	-	-	-	-	-
Cambrian-P 2026.05	89.5	-	-	74.8	-	85	73.7	-	52.6	76.9	74.9	76	-	60.1	-	-	-	-	-	-	-
GS-Reasoner (pred dep.) 2026.02	88.9	61.9	70	65.4	69.1	52.3	64.7	-	44.3	65.7	-	-	-	-	-	-	-	-	-	-	-
SpaceMind-8B 2026.06	88.4	-	-	67.2	-	70.6	69.6	-	44.3	74.2	73.3	77.3	-	61.4	-	-	-	-	-	-	-
SpaceMind 2026.06	88.4	61.4	-	67.2	-	70.6	69.6	-	44.3	74.2	73.3	77.3	-	-	-	-	-	-	-	-	-
SSR-2D 2026.02	87.9	63.4	76.5	69.2	71.8	81.9	71.9	-	48.5	76.5	-	-	-	-	-	-	-	-	-	-	-
VLM²-7B 2026.05	87.8	59.6	-	69	-	68.3	68.8	-	52.6	69.9	72.5	70.8	-	-	-	-	-	-	-	-	-
Spatial Code-4B 2026.06	87.1	60.7	-	62	-	59	60	-	32.5	33.1	92	50.8	-	-	-	-	-	-	-	-	-
Q-GeoMem 2026.05	85.5	54.5	-	71.4	-	73.4	70.1	-	46.9	76.3	76.3	76.7	-	-	-	-	-	-	-	-	-
OneCanvas 2026.06	84.8	59.5	-	71	-	65.7	70.1	-	59.8	76.5	68.4	75.4	-	-	-	-	-	-	-	-	-
SpatialStack-5B 2026.03	84.1	55.6	-	67.3	-	83.5	67.5	-	41.2	68.2	71	69.1	-	-	-	-	-	-	-	-	-
SpatialStack-5B 2026.06	84.1	-	-	67.3	-	83.5	67.5	-	41.2	68.2	71	69.1	-	55.6	-	-	-	-	-	-	-
Cambrian-S-7B† 2026.05	82	-	-	71.5	-	84.3	69.2	-	38.7	74.7	73.6	75.2	-	53.7	-	-	-	-	-	-	-
SenseNova-SI 2026.06	80.8	53.5	-	69.6	-	76.4	68.8	-	48.5	72.8	72	76.8	-	-	-	-	-	-	-	-	-
SenseNova-SI InternVL3-8B 2026.02	80.7	72	72.7	76.3	76.7	48.4	68.7	-	69.5	53.5	-	-	-	-	-	-	-	-	-	-	-
GeoVR-2B 2026.06	80.7	-	-	71.3	-	86.7	69.1	-	45.9	72.3	67.7	73.9	-	54.5	-	-	-	-	-	-	-
VLM-3R-7B 2026.02	80.5	49.4	69.2	65.4	70.2	40.1	60.9	-	45.4	67.1	-	-	-	-	-	-	-	-	-	-	-
VLM-3R (7B) 2025.05	80.5	-	-	65.4	-	40.1	60.9	-	45.4	67.1	70.2	69.2	-	49.4	-	-	-	-	-	-	-
VLM-3R-7B 2026.04	80.5	49.4	-	65.4	-	-	60.9	-	45.4	67.1	70.2	69.2	-	-	-	40.1	-	-	-	-	-
VLM-3R 7B 2026.05	80.5	-	-	65.4	-	40.1	60.9	-	45.4	67.1	70.2	69.2	-	49.4	-	-	-	-	-	-	-
VLM-3R-7B 2026.05	80.5	49.4	-	65.4	-	40.1	60.9	-	45.4	67.1	70.2	69.2	-	-	-	-	-	-	-	-	-
VLM-3R 2026.05	80.5	-	-	-	-	-	-	-	45.4	-	70.2	-	-	-	-	-	-	-	-	-	-
VLM-3R-7B 2026.06	80.5	-	-	65.4	-	40.1	60.9	-	45.4	67.1	70.2	69.2	-	49.4	-	-	-	-	-	-	-
VLM-3R-7B 2026.06	80.5	49.4	-	65.4	-	40.1	60.9	-	45.4	67.1	70.2	69.2	-	-	-	-	-	-	-	-	-
VLM-3R 2026.06	80.5	49.4	-	65.4	-	40.1	60.9	-	45.4	67.1	70.2	69.2	-	-	-	-	-	-	-	-	-
Cambrian-S-7B 2026.02	76.2	50.5	74.9	71.1	73.2	80.1	67.5	-	41.8	72.2	-	-	-	-	-	-	-	-	-	-	-
Cambrian-S 7B 2026.05	76.2	-	-	71.1	-	80.1	67.5	-	41.8	72.2	73.2	74.9	-	50.5	-	-	-	-	-	-	-
Cambrian-S-7B 2026.06	76.2	-	-	71.1	-	80.1	67.5	-	41.8	72.2	73.2	74.9	-	50.5	-	-	-	-	-	-	-
Cambrian-S 2026.06	76.2	50.5	-	71.1	-	80.1	67.5	-	41.8	72.2	73.2	74.9	-	-	-	-	-	-	-	-	-
Spatial-TTT-2B 2026.03	73	-	-	61.8	-	77	64.4	-	47.4	65.9	70.8	71.7	-	47.8	-	-	-	-	-	-	-
SpaR3D-MoE 2026.07	70.1	48	-	58.7	-	73.5	63.5	-	44	69.6	71.3	72.8	-	-	-	-	-	-	-	-	-
SpatialImaginer 2026.04	69.9	46.8	-	67.4	-	-	63.7	-	46.4	64.6	70.6	72.7	-	-	-	71.2	-	-	-	-	-
Cambrian-S-7B 2026.04	69.6	45.8	-	66.8	-	-	62.9	-	39.2	67.6	68.2	72.5	-	-	-	73.8	-	-	-	-	-
Cambrian-S-7B 2026.06	69.6	45.8	-	66.8	-	73.8	62.9	-	39.2	67.6	68.2	72.5	-	-	-	-	-	-	-	-	-
LLaVA-Next-Video-7B (Finetuned) 2025.05	68.9	-	-	64.9	-	38.5	57.7	-	40.7	63.7	70.6	70.8	-	43.6	-	-	-	-	-	-	-
SpatialStack-4B 2026.03	68.4	45.4	-	57.9	-	79.6	60.9	-	40.2	63.2	69.2	63	-	-	-	-	-	-	-	-	-
SpatialStack-4B 2026.06	68.4	-	-	57.9	-	79.6	60.9	-	40.2	63.2	69.2	63	-	45.4	-	-	-	-	-	-	-
Skill-3D 2026.06	67	-	-	62.7	-	78.1	-	-	71.4	77.6	66.2	74.9	-	61.5	-	-	-	-	-	-	-
LLaVA-One-Vision-7B (Finetuned) 2025.05	65.9	-	-	59.7	-	39.4	55.8	-	41.7	59.5	68.3	66.5	-	45.3	-	-	-	-	-	-	-
Skill-3D 2026.06	64.2	-	-	67.6	-	76.4	-	-	69	72.8	62.4	73.1	-	58	-	-	-	-	-	-	-
Ouro-Spatial-4B 2026.06	63.7	47.5	-	63.8	-	71.4	62.7	-	33	73.5	71.5	77.1	-	-	-	-	-	-	-	-	-
Skill-3D 2026.06	63	-	-	60.4	-	73.4	-	-	67.5	71.8	60.9	71.2	-	56.1	-	-	-	-	-	-	-
Ouro-Spatial-8B 2026.06	62.6	48.7	-	64.2	-	70.7	63.3	-	35.6	75.7	71.7	77.3	-	-	-	-	-	-	-	-	-
GR3D 2026.03	62.2	-	-	57.6	-	68.1	58.7	-	61.2	60.7	53.4	63.5	-	42.7	-	-	-	-	-	-	-
Cambrian-S-3B 2026.03	61.9	-	-	64.8	-	78.8	57.3	-	27.3	46.3	70.7	68	-	40.6	-	-	-	-	-	-	-
Cambrian-S-3B 2026.03	61.9	40.6	-	64.8	-	78.8	57.3	-	27.3	46.3	70.7	68	-	-	-	-	-	-	-	-	-
Cambrian-S-3B 2026.06	61.9	-	-	64.8	-	78.8	57.3	-	27.3	46.3	70.7	68	-	40.6	-	-	-	-	-	-	-
Cambrian-S-3B 2026.06	61.9	40.6	-	64.8	-	78.8	57.3	-	27.3	46.3	70.7	68	-	-	-	-	-	-	-	-	-
Think3D 2026.06	61.8	-	-	58.3	-	73.8	-	-	66.5	69.9	61.1	70.8	-	55	-	-	-	-	-	-	-
Think3D 2026.06	60.8	-	-	56.5	-	69.3	-	-	64.2	66.3	56.8	68	-	52.3	-	-	-	-	-	-	-
Skill-3D-8B 2026.06	60.1	48.6	-	52	-	66.8	-	-	58.4	67.9	56.5	59.8	-	-	-	-	-	-	-	-	-
Think3D 2026.06	59.5	-	-	64.8	-	72.2	-	-	65.8	66.4	58.2	69.5	-	53.1	-	-	-	-	-	-	-
Skill-3D 2026.06	59.2	-	-	53.4	-	35.4	-	-	62.7	69.5	56.8	58.1	-	42.6	-	-	-	-	-	-	-
Gemini-3-pro 2026.03	57.5	-	-	56.6	-	60	56	-	61.9	41.8	49	71.5	-	42.8	-	-	-	-	-	-	-
Qwen3.5-4B 2026.03	57.5	36.5	-	60.3	-	62.3	53.6	-	34	53.8	56.5	67.5	-	-	-	-	-	-	-	-	-
Qwen3.5-4B 2026.06	57.5	-	-	60.3	-	62.3	53.6	-	34	53.8	56.5	67.5	-	36.5	-	-	-	-	-	-	-
Gemini-3-Pro 2026.06	57.5	42.8	-	56.6	-	68	56	-	61.9	41.8	49	71.5	-	-	-	-	-	-	-	-	-
ReMoT -CoT-4B 2026.02	57.4	42.5	-	52.1	-	59.7	58.8	-	45.6	56.3	78.2	76.8	1	-	-	-	-	-	-	-	-
VST-7B-SFT 2026.03	55.8	-	-	59.7	-	65.2	60.6	-	44.9	68.3	72	74.3	-	44.4	-	-	-	-	-	-	-
VST-7B 2026.05	55.8	44.4	-	59.7	-	65.2	60.6	-	44.9	68.3	72	74.3	-	-	-	-	-	-	-	-	-
VST-7B-SFT 2026.06	55.8	-	-	59.7	-	65.2	60.6	-	44.9	68.3	72	74.3	-	44.4	-	-	-	-	-	-	-
VST-7B-SFT 2026.06	55.8	44.4	-	59.7	-	65.2	60.6	-	44.9	68.3	72	74.3	-	-	-	-	-	-	-	-	-
Gemini-3-Pro 2026.04	55.7	37.8	-	59.9	-	-	52.5	-	45.9	44.1	38	72.7	-	-	-	66	-	-	-	-	-
VST-7B 2026.02	55.6	43.8	75.5	60	71.6	69.2	61.2	-	44.3	69.2	-	-	-	-	-	-	-	-	-	-	-
VST 7B 2026.05	55.6	-	-	60	-	69.2	61.2	-	44.3	69.2	71.6	75.5	-	43.8	-	-	-	-	-	-	-
VST 2026.06	55.6	43.8	-	60	-	69.2	61.2	-	44.3	69.2	71.6	75.5	-	-	-	-	-	-	-	-	-
w/ Tools 2026.06	53.3	-	-	51.7	-	74.6	-	-	52.5	58.2	58	69.3	-	47.8	-	-	-	-	-	-	-
Think3D-8B 2026.06	52.9	38.5	-	41.6	-	60.8	-	-	45.4	58.1	48.3	51.2	-	-	-	-	-	-	-	-	-
Think3D 2026.06	52.6	-	-	47.9	-	29.2	-	-	55.9	61.4	50.4	50.5	-	32.7	-	-	-	-	-	-	-
Qwen3-VL-8B 2026.06	51.5	47.1	-	58.3	-	64.2	56.5	-	31.4	57.2	66	75.9	-	-	-	-	-	-	-	-	-
w/ Tools 2026.06	51.2	-	-	62.6	-	72.7	-	-	49.5	46.3	48.4	66	-	41.5	-	-	-	-	-	-	-
Qwen3-VL-8B-Instruct 2026.04	51	47	-	58	-	-	57.9	-	35	61.9	67.6	76.3	-	-	-	66.3	-	-	-	-	-
Qwen3-VL-8B-Instruct 2026.02	50.9	47	76.3	58	53.3	66.3	57.9	-	35	61.9	-	-	-	-	-	-	-	-	-	-	-
Qwen3-VL-8B-Instruct 2026.03	50.9	-	-	58	-	66.3	57.9	-	35	61.9	67.5	76.3	-	47	-	-	-	-	-	-	-
Qwen3-VL-8B-Instruct 2026.06	50.9	-	-	58	-	66.3	57.9	-	35	61.9	67.5	76.3	-	47	-	-	-	-	-	-	-
w/o Tools 2026.06	50.5	-	-	48.4	-	73.2	-	-	49.1	55.7	55.8	67.3	-	43.6	-	-	-	-	-	-	-
Skill-3D-4B 2026.06	50.4	36.8	-	43.5	-	56.7	-	-	48.8	57.4	48.6	50.2	-	-	-	-	-	-	-	-	-
Robust-TO (Qwen3-VL-7B-Instruct) 2026.06	48.8	-	-	55.5	-	-	56.4	-	39.8	-	-	-	-	-	-	-	-	-	-	-	77.5
GPT-5 2026.04	48.7	34.5	-	63.7	-	-	55	-	50.3	47.5	53.3	73.3	-	-	-	68.9	-	-	-	-	-
VST-7B-SFT 2026.04	48.7	37.3	-	55.2	-	-	55.5	-	41.8	62.2	68.8	74.5	-	-	-	55.5	-	-	-	-	-
GPT-5-2025-08-07 2026.02	48.6	34.4	73.3	63.7	53.5	68.9	55	-	50.2	47.5	-	-	-	-	-	-	-	-	-	-	-
GPT-5 2026.03	48.6	-	-	63.7	-	68.9	55	-	50.2	47.5	53.3	73.3	-	34.4	-	-	-	-	-	-	-
GPT-5 2026.06	48.6	-	-	63.7	-	68.9	55	-	50.2	47.5	53.3	73.3	-	34.4	-	-	-	-	-	-	-
GPT-5 2026.06	48.6	34.4	-	47.8	-	68.9	55	-	50.2	48.3	53.3	73.3	-	-	-	-	-	-	-	-	-
w/ Tools 2026.06	48.4	27.8	-	44.7	-	52.6	-	-	43.2	46	44.7	48.1	-	-	-	-	-	-	-	-	-
Spatial-MLLM-4B 2026.04	48.2	38.1	-	40.4	-	-	46.3	-	33	35.5	66.7	63.6	-	-	-	44.3	-	-	-	-	-
Gemini-1.5-Pro 2026.03	48.1	28.8	-	46	-	68	48.8	-	42	49.4	49.6	58.6	-	-	-	-	-	-	-	-	-
Random Choice 2026.03	47.9	-	-	25.1	-	25.2	34	-	28.4	33.1	62.1	29.9	-	32	-	-	-	-	-	-	-
Chance Level (Frequency) 2026.03	47.9	32	-	25.1	-	25.2	34	-	28.4	33.1	62.1	29.9	-	-	-	-	-	-	-	-	-
Chance Level (Frequency) 2025.05	47.9	-	-	25.1	-	25.2	34	-	28.4	33.1	62.1	29.9	-	32	-	-	-	-	-	-	-
Random Choice(Frequency) 2026.04	47.9	32	-	25.1	-	-	34	-	28.4	33.1	62.1	29.9	-	-	-	25.2	-	-	-	-	-
Chance Level (Frequency) 2026.05	47.9	-	-	25.1	-	25.2	34	-	28.4	33.1	62.1	29.9	-	32	-	-	-	-	-	-	-
Random 2026.06	47.9	32	-	25.1	-	25.2	34	-	28.4	33.1	62.1	29.9	-	-	-	-	-	-	-	-	-
Gemini-2.5 Pro 2026.03	47.8	34.9	-	61.1	-	71.3	51.5	-	45.9	42.8	43.8	64.3	-	-	-	-	-	-	-	-	-
Gemini-2.5 Pro 2025.05	47.8	-	-	61.1	-	71.3	51.5	-	45.9	42.8	43.8	64.3	-	34.9	-	-	-	-	-	-	-

Spatial Reasoning on VSI-Bench

Evaluation Results