# Long-horizon language-conditioned manipulation on Calvin ABC→D

Source: [Long-horizon language-conditioned manipulation on Calvin ABC→D](<https://www.sota2.com/research/sota/long-horizon-language-conditioned-manipulation-on-calvin-abc-d>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-04T07:05:29.781Z

Evaluation of model performance on the Calvin ABC→D dataset for long-horizon, language-conditioned manipulation, measured by the success rate of completing 1 through 5 tasks sequentially and the average length of successful sequences.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Long-horizon language-conditioned manipulation |
| Dataset (including any stated split) | Calvin ABC→D |
| Metric | Average Sequence Length |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 18 of 18 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Average Sequence Length](<https://www.sota2.com/research/sota/long-horizon-language-conditioned-manipulation-on-calvin-abc-d.md?metric=Average%20Sequence%20Length>): Higher is better; unit: SCALAR.

- [Success Rate (Seq 1)](<https://www.sota2.com/research/sota/long-horizon-language-conditioned-manipulation-on-calvin-abc-d.md?metric=Success%20Rate%20(Seq%201)>): Higher is better; unit: PERCENT.

- [Success Rate (Seq 2)](<https://www.sota2.com/research/sota/long-horizon-language-conditioned-manipulation-on-calvin-abc-d.md?metric=Success%20Rate%20(Seq%202)>): Higher is better; unit: PERCENT.

- [Success Rate (Seq 3)](<https://www.sota2.com/research/sota/long-horizon-language-conditioned-manipulation-on-calvin-abc-d.md?metric=Success%20Rate%20(Seq%203)>): Higher is better; unit: PERCENT.

- [Success Rate (Seq 4)](<https://www.sota2.com/research/sota/long-horizon-language-conditioned-manipulation-on-calvin-abc-d.md?metric=Success%20Rate%20(Seq%204)>): Higher is better; unit: PERCENT.

- [Success Rate (Seq 5)](<https://www.sota2.com/research/sota/long-horizon-language-conditioned-manipulation-on-calvin-abc-d.md?metric=Success%20Rate%20(Seq%205)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Average Sequence Length | Result date | Report paper |
| --- | --- | --- | --- |
| X-VLA | 4.43 | Not recorded | 6a21067bc10484bac23efff5 |
| VLA-Adapter | 4.42 | Not recorded | 6a21067bc10484bac23efff5 |
| UniJEPA | 4.11 | Not recorded | 68fa7c0b1e209131cea8c0d9 |
| UP-VLA\* | 4.08 | Not recorded | 68fa7c0b1e209131cea8c0d9 |
| S2 | 3.95 | Not recorded | 6a21067bc10484bac23efff5 |
| π0.5 | 3.92 | Not recorded | 6a21067bc10484bac23efff5 |
| UniVLA | 3.8 | Not recorded | 6a21067bc10484bac23efff5 |
| Lie Diffuser Actor | 3.584 | Not recorded | 6a1e6438dbd7150506f85881 |
| VPP\* | 3.58 | Not recorded | 68fa7c0b1e209131cea8c0d9 |
| Lie Diffuser Actor | 3.512 | Not recorded | 6a1e6438dbd7150506f85881 |
| Lie Diffuser Actor | 3.474 | Not recorded | 6a1e6438dbd7150506f85881 |
| Lie Diffuser Actor | 3.404 | Not recorded | 6a1e6438dbd7150506f85881 |
| Lie Diffuser Actor | 3.368 | Not recorded | 6a1e6438dbd7150506f85881 |
| 3D Diffuser Actor | 3.288 | Not recorded | 6a1e6438dbd7150506f85881 |
| 3D Diffuser Actor | 3.27 | Not recorded | 6a1e6438dbd7150506f85881 |
| OpenVLA-OFT | 3.27 | Not recorded | 6a21067bc10484bac23efff5 |
| Lie Diffuser Actor | 3.239 | Not recorded | 6a1e6438dbd7150506f85881 |
| GR-1 | 3.06 | Not recorded | 68fa7c0b1e209131cea8c0d9 |

### X-VLA

Average Sequence Length: 4.43. Result ID: 6a2123b6fbc373592c1c25b2.

[Reporting paper](<https://www.sota2.com/research/paper/6a21067bc10484bac23efff5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c091e209131cea8c027.md>)

Recorded evaluation configuration: {"Suite":"CALVIN ABC→D"}. This record does not assert independent reproduction.

### VLA-Adapter

Average Sequence Length: 4.42. Result ID: 6a2123b6fbc373592c1c25b1.

[Reporting paper](<https://www.sota2.com/research/paper/6a21067bc10484bac23efff5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b861e209131cea88e51.md>)

Recorded evaluation configuration: {"Suite":"CALVIN ABC→D"}. This record does not assert independent reproduction.

### UniJEPA

Average Sequence Length: 4.11. Result ID: 6a056f7b47f34ce924ebc9fd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0b1e209131cea8c0d9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c0b1e209131cea8c0d9.md>)

Recorded evaluation configuration: {"Visual prediction":"true","Input resolution":"224x224","Input view":"single third-view"}. This record does not assert independent reproduction.

### UP-VLA\*

Average Sequence Length: 4.08. Result ID: 6a056f7b47f34ce924ebc9fc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0b1e209131cea8c0d9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79a11e209131cea7e8d4.md>)

[Implementation](<https://github.com/CladernyJorn/UP-VLA>)

Recorded evaluation configuration: {"Visual prediction":"true","Input resolution":"224x224","Input view":"single third-view","Reproduction status":"reproduced"}. This record does not assert independent reproduction.

### S2

Average Sequence Length: 3.95. Result ID: 6a2123b6fbc373592c1c25b4.

[Reporting paper](<https://www.sota2.com/research/paper/6a21067bc10484bac23efff5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a21067bc10484bac23efff5.md>)

Recorded evaluation configuration: {"Suite":"CALVIN ABC→D","interface":"Specify More language interface"}. This record does not assert independent reproduction.

### π0.5

Average Sequence Length: 3.92. Result ID: 6a2123b6fbc373592c1c25b3.

[Reporting paper](<https://www.sota2.com/research/paper/6a21067bc10484bac23efff5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa783a1e209131cea774f1.md>)

Recorded evaluation configuration: {"Suite":"CALVIN ABC→D"}. This record does not assert independent reproduction.

### UniVLA

Average Sequence Length: 3.8. Result ID: 6a2123b6fbc373592c1c25b0.

[Reporting paper](<https://www.sota2.com/research/paper/6a21067bc10484bac23efff5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79ab1e209131cea7ecf5.md>)

Recorded evaluation configuration: {"Suite":"CALVIN ABC→D"}. This record does not assert independent reproduction.

### Lie Diffuser Actor

Average Sequence Length: 3.584. Result ID: 6a1eca2afbc373592c1a68ca.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

Recorded evaluation configuration: {"Training Iterations":"300K"}. This record does not assert independent reproduction.

### VPP\*

Average Sequence Length: 3.58. Result ID: 6a056f7b47f34ce924ebc9fb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0b1e209131cea8c0d9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78581e209131cea77da7.md>)

[Implementation](<https://video-prediction-policy.github.io/>)

Recorded evaluation configuration: {"Visual prediction":"true","Input resolution":"224x224","Input view":"single third-view","Reproduction status":"reproduced"}. This record does not assert independent reproduction.

### Lie Diffuser Actor

Average Sequence Length: 3.512. Result ID: 6a1eca2afbc373592c1a68c6.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

Recorded evaluation configuration: {"Training Iterations":"300K"}. This record does not assert independent reproduction.

### Lie Diffuser Actor

Average Sequence Length: 3.474. Result ID: 6a1eca2afbc373592c1a68c5.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

Recorded evaluation configuration: {"Training Iterations":"300K","Lie Space Diffusion":"excluded"}. This record does not assert independent reproduction.

### Lie Diffuser Actor

Average Sequence Length: 3.404. Result ID: 6a1eca2afbc373592c1a68c8.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

Recorded evaluation configuration: {"Training Iterations":"300K","GAT Encoder":"excluded"}. This record does not assert independent reproduction.

### Lie Diffuser Actor

Average Sequence Length: 3.368. Result ID: 6a1eca2afbc373592c1a68c4.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

Recorded evaluation configuration: {"Training Iterations":"600K","GAT Encoder":"excluded"}. This record does not assert independent reproduction.

### 3D Diffuser Actor

Average Sequence Length: 3.288. Result ID: 6a1eca2afbc373592c1a68c7.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa744b1e209131cea60daf.md>)

[Implementation](<https://www.sota2.com/3d-diffuser-actor.github.io>)

Recorded evaluation configuration: {"Training Iterations":"800K"}. This record does not assert independent reproduction.

### 3D Diffuser Actor

Average Sequence Length: 3.27. Result ID: 6a1eca2afbc373592c1a68c3.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa744b1e209131cea60daf.md>)

[Implementation](<https://www.sota2.com/3d-diffuser-actor.github.io>)

Recorded evaluation configuration: {"Training Iterations":"600K"}. This record does not assert independent reproduction.

### OpenVLA-OFT

Average Sequence Length: 3.27. Result ID: 6a2123b6fbc373592c1c25af.

[Reporting paper](<https://www.sota2.com/research/paper/6a21067bc10484bac23efff5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78471e209131cea7794b.md>)

Recorded evaluation configuration: {"Suite":"CALVIN ABC→D"}. This record does not assert independent reproduction.

### Lie Diffuser Actor

Average Sequence Length: 3.239. Result ID: 6a1eca2afbc373592c1a68c9.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1e6438dbd7150506f85881.md>)

Recorded evaluation configuration: {"Training Iterations":"400K","Lie Space Diffusion":"excluded"}. This record does not assert independent reproduction.

### GR-1

Average Sequence Length: 3.06. Result ID: 6a056f7b47f34ce924ebc9fa.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0b1e209131cea8c0d9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71861e209131cea51c7f.md>)

[Implementation](<https://gr1-manipulation.github.io/>)

Recorded evaluation configuration: {"Visual prediction":"true","Input resolution":"224x224","Input view":"single third-view"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.
