Language modelOpen weights

ZDTaichu5.0-9B

Taichu AI

Released: -
Data date: October 3, 2026

ZDTaichu5.0-9B adds a C-RADIOv4-H vision encoder to Qwen3.5-9B. Taichu AI targets spatial reasoning with images and videos through this combination. The multimodal checkpoint handles up to 128K context tokens.

The published weights use the NVIDIA Open Model License Agreement. The model card also retains Qwen3.5’s Apache 2.0 license and notices for other third-party components.

Specifications and access

Specification	Value and source
Model ID	TaichuAI/ZDTaichu5.0-9BSource
Model class	Multimodal causal language model with vision encoderSource
Architecture	Qwen3.5-9B with a C-RADIOv4-H vision encoderSource
Input	Text, images, and videoSource
Context window	Up to 128K tokens according to the model cardSource

Measurements without matching peer values

These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.

CV-Bench (Show measurement, test conditions, and source)

Source value: 86.82
Score: 86.82
Metric: CV-Bench reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Vendor model card; agent tasks use a simulator/judge and several spatial tasks use an extra thinking/boxed-output prompt. Conflicting public vendor figures exist for some spatial rows.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

3DSRBench (Show measurement, test conditions, and source)

Source value: 60.96
Score: 60.96
Metric: 3DSRBench reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Vendor model card; agent tasks use a simulator/judge and several spatial tasks use an extra thinking/boxed-output prompt. Conflicting public vendor figures exist for some spatial rows.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

Spar (Show measurement, test conditions, and source)

Source value: 51.82
Score: 51.82
Metric: Spar reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Vendor model card; agent tasks use a simulator/judge and several spatial tasks use an extra thinking/boxed-output prompt. Conflicting public vendor figures exist for some spatial rows.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

ViewSpatial (Show measurement, test conditions, and source)

Source value: 62.5
Score: 62.5
Metric: ViewSpatial reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Vendor model card; agent tasks use a simulator/judge and several spatial tasks use an extra thinking/boxed-output prompt. Conflicting public vendor figures exist for some spatial rows.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

MMSI (Show measurement, test conditions, and source)

Source value: 47.2
Score: 47.2
Metric: MMSI reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Vendor model card; agent tasks use a simulator/judge and several spatial tasks use an extra thinking/boxed-output prompt. Conflicting public vendor figures exist for some spatial rows.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

MindCube (Show measurement, test conditions, and source)

Source value: 78.27
Score: 78.27
Metric: MindCube reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Vendor model card; agent tasks use a simulator/judge and several spatial tasks use an extra thinking/boxed-output prompt. Conflicting public vendor figures exist for some spatial rows.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

RoboSpatial (Show measurement, test conditions, and source)

Source value: 56
Score: 56
Metric: RoboSpatial reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Vendor model card; agent tasks use a simulator/judge and several spatial tasks use an extra thinking/boxed-output prompt. Conflicting public vendor figures exist for some spatial rows.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

MMLU-Pro (Show measurement, test conditions, and source)

Source value: 77.2
Score: 77.2
Metric: MMLU-Pro reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Vendor model card; agent tasks use a simulator/judge and several spatial tasks use an extra thinking/boxed-output prompt. Conflicting public vendor figures exist for some spatial rows.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

IFEval (Show measurement, test conditions, and source)

Source value: 93.7
Score: 93.7
Metric: IFEval reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Vendor model card; agent tasks use a simulator/judge and several spatial tasks use an extra thinking/boxed-output prompt. Conflicting public vendor figures exist for some spatial rows.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

HMMT Feb 2026 (Show measurement, test conditions, and source)

Source value: 72.7
Score: 72.7
Metric: HMMT Feb 2026 reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Vendor model card; agent tasks use a simulator/judge and several spatial tasks use an extra thinking/boxed-output prompt. Conflicting public vendor figures exist for some spatial rows.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

LiveCodeBench v6 (Show measurement, test conditions, and source)

Source value: 73.4
Score: 73.4
Metric: LiveCodeBench v6 reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Vendor model card; agent tasks use a simulator/judge and several spatial tasks use an extra thinking/boxed-output prompt. Conflicting public vendor figures exist for some spatial rows.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

tau2-bench (Show measurement, test conditions, and source)

Source value: 87.7
Score: 87.7
Metric: tau2-bench reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Vendor model card; agent tasks use a simulator/judge and several spatial tasks use an extra thinking/boxed-output prompt. Conflicting public vendor figures exist for some spatial rows.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

Page 1 of 2

Sources and data date

Every statement links to its underlying documentation or leaderboard.

Type	Evidence and data date
Research status	Research date October 4, 2026. 1 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication.
Additional source	ZDTaichu5.0-9B model card (retrieved October 3, 2026; October 4, 2026) · ZDTaichu5.0-9B model card · Editorial description reviewed October 4, 2026