# Text-to-TikZ on SPIQA (test-A and test-B)

Source: [Text-to-TikZ on SPIQA (test-A and test-B)](<https://www.sota2.com/research/sota/text-to-tikz-on-spiqa-test-a-and-test-b>)

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

Data updated: 2026-03-04T20:51:53.597Z

Evaluation of Text-to-TikZ models on the SPIQA dataset, assessing performance across the test-A and test-B splits using CLIP, DSim, and CR metrics.

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 | Text-to-TikZ |
| Dataset (including any stated split) | SPIQA (test-A and test-B) |
| Metric | CLIP Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 9 of 9 |
| 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

- [CLIP Score](<https://www.sota2.com/research/sota/text-to-tikz-on-spiqa-test-a-and-test-b.md?metric=CLIP%20Score>): Higher is better; unit: SCALAR.

- [DSim Score](<https://www.sota2.com/research/sota/text-to-tikz-on-spiqa-test-a-and-test-b.md?metric=DSim%20Score>): Higher is better; unit: SCALAR.

- [CR Rate](<https://www.sota2.com/research/sota/text-to-tikz-on-spiqa-test-a-and-test-b.md?metric=CR%20Rate>): Higher is better; unit: PERCENT.

## Recorded results

| Method | CLIP Score | Result date | Report paper |
| --- | --- | --- | --- |
| TikZilla-3B-RL | 0.193 | Not recorded | 69a7bc852d603e6de3e3d541 |
| TikZilla-8B-RL | 0.174 | Not recorded | 69a7bc852d603e6de3e3d541 |
| TikZilla-8B | 0.131 | Not recorded | 69a7bc852d603e6de3e3d541 |
| GPT-5 | 0.115 | Not recorded | 69a7bc852d603e6de3e3d541 |
| TikZilla-3B | 0.114 | Not recorded | 69a7bc852d603e6de3e3d541 |
| Qwen3-Coder-30B-A3B | 0.102 | Not recorded | 69a7bc852d603e6de3e3d541 |
| GPT-4o | 0.098 | Not recorded | 69a7bc852d603e6de3e3d541 |
| Qwen3-8B | 0.07 | Not recorded | 69a7bc852d603e6de3e3d541 |
| Qwen2.5-3B | 0.038 | Not recorded | 69a7bc852d603e6de3e3d541 |

### TikZilla-3B-RL

CLIP Score: 0.193. Result ID: 69a89b636dbb7075b9517e64.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

Recorded evaluation configuration: {"AT":"765"}. This record does not assert independent reproduction.

### TikZilla-8B-RL

CLIP Score: 0.174. Result ID: 69a89b636dbb7075b9517e67.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

Recorded evaluation configuration: {"AT":"809"}. This record does not assert independent reproduction.

### TikZilla-8B

CLIP Score: 0.131. Result ID: 69a89b636dbb7075b9517e66.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

Recorded evaluation configuration: {"AT":"1184"}. This record does not assert independent reproduction.

### GPT-5

CLIP Score: 0.115. Result ID: 69a89b636dbb7075b9517e5f.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

Recorded evaluation configuration: {"AT":"1239"}. This record does not assert independent reproduction.

### TikZilla-3B

CLIP Score: 0.114. Result ID: 69a89b636dbb7075b9517e63.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

Recorded evaluation configuration: {"AT":"1170"}. This record does not assert independent reproduction.

### Qwen3-Coder-30B-A3B

CLIP Score: 0.102. Result ID: 69a89b636dbb7075b9517e61.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa788e1e209131cea79116.md>)

Recorded evaluation configuration: {"AT":"1000"}. This record does not assert independent reproduction.

### GPT-4o

CLIP Score: 0.098. Result ID: 69a89b636dbb7075b9517e60.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75681e209131cea67a98.md>)

Recorded evaluation configuration: {"AT":"748"}. This record does not assert independent reproduction.

### Qwen3-8B

CLIP Score: 0.07. Result ID: 69a89b636dbb7075b9517e65.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa788e1e209131cea79116.md>)

Recorded evaluation configuration: {"AT":"781"}. This record does not assert independent reproduction.

### Qwen2.5-3B

CLIP Score: 0.038. Result ID: 69a89b636dbb7075b9517e62.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc852d603e6de3e3d541.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa767e1e209131cea6d6a7.md>)

Recorded evaluation configuration: {"AT":"772"}. This record does not assert independent reproduction.

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