# Multi-Token Prediction on SPEED-Bench Throughput-1k

Source: [Multi-Token Prediction on SPEED-Bench Throughput-1k](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k>)

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

Data updated: 2026-04-15T08:35:55.393Z

Evaluation of model performance across various capability domains (Coding, Math, Reasoning, etc.) using the SPEED-Bench Throughput-1k dataset, likely focusing on throughput efficiency during multi-token generation.

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 | Multi-Token Prediction |
| Dataset (including any stated split) | SPEED-Bench Throughput-1k |
| Metric | Coding Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 3 of 3 |
| 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

- [Coding Score](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k.md?metric=Coding%20Score>): Higher is better; unit: SCALAR.

- [Humanities Score](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k.md?metric=Humanities%20Score>): Higher is better; unit: SCALAR.

- [Math Score](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k.md?metric=Math%20Score>): Higher is better; unit: SCALAR.

- [Multilingual Score](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k.md?metric=Multilingual%20Score>): Higher is better; unit: SCALAR.

- [QA Score](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k.md?metric=QA%20Score>): Higher is better; unit: SCALAR.

- [RAG Score](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k.md?metric=RAG%20Score>): Higher is better; unit: SCALAR.

- [Reasoning Score](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k.md?metric=Reasoning%20Score>): Higher is better; unit: SCALAR.

- [Roleplay Score](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k.md?metric=Roleplay%20Score>): Higher is better; unit: SCALAR.

- [STEM Score](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k.md?metric=STEM%20Score>): Higher is better; unit: SCALAR.

- [Summarization Score](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k.md?metric=Summarization%20Score>): Higher is better; unit: SCALAR.

- [Writing Score](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k.md?metric=Writing%20Score>): Higher is better; unit: SCALAR.

- [Average Score](<https://www.sota2.com/research/sota/multi-token-prediction-on-speed-bench-throughput-1k.md?metric=Average%20Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Coding Score | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen3-Next | 4.32 | Not recorded | 69df1b757d0172140ec7b7f9 |
| Nemotron 3 Super | 3.78 | Not recorded | 69df1b757d0172140ec7b7f9 |
| DeepSeek-R1 | 2.99 | Not recorded | 69df1b757d0172140ec7b7f9 |

### Qwen3-Next

Coding Score: 4.32. Result ID: 69df4ddd47f34ce924c815ef.

[Reporting paper](<https://www.sota2.com/research/paper/69df1b757d0172140ec7b7f9.md>)

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

Recorded evaluation configuration: {"Draft Length":"7"}. This record does not assert independent reproduction.

### Nemotron 3 Super

Coding Score: 3.78. Result ID: 69df4ddd47f34ce924c815f0.

[Reporting paper](<https://www.sota2.com/research/paper/69df1b757d0172140ec7b7f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69df1b757d0172140ec7b7f9.md>)

Recorded evaluation configuration: {"Draft Length":"7"}. This record does not assert independent reproduction.

### DeepSeek-R1

Coding Score: 2.99. Result ID: 69df4ddd47f34ce924c815ee.

[Reporting paper](<https://www.sota2.com/research/paper/69df1b757d0172140ec7b7f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76a91e209131cea6e3eb.md>)

Recorded evaluation configuration: {"Draft Length":"7"}. This record does not assert independent reproduction.

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