Help Center

Models

Bugfixes / Misc

Features

Token

Characters

veniceai

New Submission

Backlog

In Review

Planned

Up next

In Progress

Completed

Rejected

High Priority

Low Priority

New Model

Chat

History

Image

Support

Web Search

Backup / Sync

Staking Contract

Voice

Points

Settings

Code

Folders

Payment

Video

UI Design

Stickers

Merchandise

Censorship

Context Window

Social

Enhance Prompt

Plugin

File Upload

Upscale/Enhance

Edit Prompt

Language

Privacy

Venice Pro

System Prompts

Vision

LoRAs

Image Editing

Memory

Social Feed

Artifacts

Auto Mode

Blog

Incentive Fund

Agents

Model Deprecation

Credits

Library

Projects

Marketing

Here’s what Venice is working on next!

Next up

Roadmap

All completed and shipped community requested feature requests and improvements can be seen here.

Done

Shipped Features

Hey {name|there}! 👋

I’d like to request the adoption of Multi-Token Prediction (MTP drafter for Gemma-4-31b, which can boost its throughput up to 3x without any loss in performance. It would be a game changer for a variety of use cases.

Ich möchte die Einführung von „Multi-Token Prediction“ (MTP-Drafter für Gemma-4-31b) beantragen, wodurch sich der Durchsatz ohne Leistungseinbußen um das Dreifache steigern lässt. Dies wäre für eine Vielzahl von Anwendungsfällen ein entscheidender Fortschritt.

Entwurf für die Multi-Token-Vorhersage (MTP) für Gemma-4-31b

Me gustaría solicitar la adopción de la predicción multitoken (MTP), propuesta para Gemma-4-31b, que puede triplicar su rendimiento sin ninguna pérdida de calidad. Supondría un gran avance para una gran variedad de casos de uso.

Generador de predicciones multitoken (MTP) para Gemma-4-31b

Multi-Token Prediction (MTP) drafter for Gemma-4-31b

Dongwon

Venice.ai

Multi-Token Prediction (MTP) drafter for Gemma-4-31b

Subscribe to post

Subscribe to post