Gemma 4 Technical Report

Introducing Gemma 4 12B: a unified, encoder-free multimodal model

TTE-Flash: Accelerating Reasoning-based Multimodal Representations via Think-Then-Embed Tokens

arXiv:2605.16638v2 Announce Type: replace Abstract: Recent research has demonstrated that Universal Multimodal Embedding (UME) benefits significantly from Chain-of-Thought (CoT) reasoning. In this pa...

M*: A Modular, Extensible, Serving System for Multimodal Models

arXiv:2606. 12688v1 Announce Type: cross Abstract: We are entering a new era of composite model architectures that integrate diverse components such as vision encoders, language backbones, diffusion and flow heads, audio codecs, action generators, and world-model predictors.

IronLLM: Forging Compact Edge-Native Language Models for Real-Time Embodied Intelligence

IronLLM-0.6B is a 654‑million‑parameter language model engineered for efficient on‑device inference, featuring a hybrid attention architecture, X‑MTP multi‑token prediction, and a lightweight verification head that yields a 1.48× decoding speedup. Trained on roughly 6.2 trillion tokens with a quality‑oriented pipeline and further refined via Multi‑Domain On‑Policy Distillation, the model adopts an Instruct‑Only design to meet low‑latency requirements. A lighter variant, IronLLM‑0.6B‑Light, replaces RMSNorm with Dynamic Tanh and streamlines costly components to enhance inference and quantization efficiency, offering a strong performance‑efficiency trade‑off for resource‑constrained deployment.

Does Understanding Inform Generation in Unified Multimodal Models? From Analysis to Path Forward

The paper introduces UniSandbox, a decoupled evaluation framework with controlled synthetic datasets, to study whether understanding informs generation in Unified Multimodal Models. Results show a notable understanding‑generation gap, especially in reasoning generation and knowledge transfer. Explicit Chain‑of‑Thought (CoT) in the understanding module bridges this gap, and self‑training can internalize CoT for implicit reasoning during generation; query‑based architectures also exhibit latent CoT‑like properties that aid knowledge transfer.

RetroThinker: Enabling Retrospective Thinking in Speech LLMs

RetroThinker is a multi-stage post‑training framework that enhances SpeechLLMs by enabling them to self‑verify and forward‑correct Chain‑of‑Thought reasoning steps during inference. It combines supervised fine‑tuning on curated retrospective thinking data with length‑based direct preference optimization to improve reasoning while the user speaks. On the GSM8K benchmark, RetroThinker achieves an 11% absolute accuracy gain over non‑retrospective baselines while maintaining comparable latency.

When Scaling Fails: Mitigating Audio Perception Decay of LALMs via Multi-Step Perception-Aware Reasoning

arXiv:2603.02266v2 Announce Type: replace-cross Abstract: Test-Time Scaling has shown notable efficacy in addressing complex problems through scaling inference compute. However, within Large Audio-La...

UniAE-MoE: A Unified Audio Encoder via Mixture of Experts

UniAE-MoE is a unified audio encoder that uses a Mixture-of-Experts architecture to model cross‑domain audio representations. It integrates encoder components from Qwen2‑Audio and Audio‑Flamingo 3, enhances them with SwiGLU and shared experts, and applies a two‑stage instruction‑tuning strategy along with task‑specific data scaling. The model achieves state‑of‑the‑art results on the XARES‑LLM benchmark (0.802) and tops the Interspeech 2026 Audio Encoder Capability Challenge, demonstrating strong generalization across speech, music, and general audio tasks.

MACS: Modality-Aware Capacity Scaling for Efficient Multimodal MoE Inference

arXiv:2605. 05225v3 Announce Type: replace-cross Abstract: Mixture-of-Experts Multimodal Large Language Models (MoE MLLMs) suffer from a significant efficiency bottleneck during Expert Parallelism (EP) inference due to the straggler effect.

Selection-Based Structured Reasoning: Toward Efficient Multimodal Search Agents

arXiv:2610.01892v1 Announce Type: cross Abstract: Multimodal agents commonly generate free-form reasoning before each action. For small models, limited model capacity can result in lengthy reasoning...

Omni2LoRA: Coherence-Preserving Parametric Memory for Efficient Omni Language Models

arXiv:2608. 09227v1 Announce Type: new Abstract: Omnimodal language models (OLMs) enable unified audio-visual understanding, but processing long joint token sequences makes inference computationally prohibitive.

VideoMM: Adaptive Macro-Micro Inference for Efficient Video MLLMs

arXiv:2609.16722v1 Announce Type: new Abstract: Scaling Multimodal Large Language Models (MLLMs) to long-form video understanding is bottlenecked by the explosion of visual tokens, which saturates co...