REVIEW 57 cited by
MiMo-VL Technical Report
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
MiMo-VL Technical Report
read the original abstract
We open-source MiMo-VL-7B-SFT and MiMo-VL-7B-RL, two powerful vision-language models delivering state-of-the-art performance in both general visual understanding and multimodal reasoning. MiMo-VL-7B-RL outperforms Qwen2.5-VL-7B on 35 out of 40 evaluated tasks, and scores 59.4 on OlympiadBench, surpassing models with up to 78B parameters. For GUI grounding applications, it sets a new standard with 56.1 on OSWorld-G, even outperforming specialized models such as UI-TARS. Our training combines four-stage pre-training (2.4 trillion tokens) with Mixed On-policy Reinforcement Learning (MORL) integrating diverse reward signals. We identify the importance of incorporating high-quality reasoning data with long Chain-of-Thought into pre-training stages, and the benefits of mixed RL despite challenges in simultaneous multi-domain optimization. We also contribute a comprehensive evaluation suite covering 50+ tasks to promote reproducibility and advance the field. The model checkpoints and full evaluation suite are available at https://github.com/XiaomiMiMo/MiMo-VL.
Forward citations
Cited by 57 Pith papers
-
DataComp-VLM: Improved Open Datasets for Vision-Language Models
DataComp-VLM benchmark shows instruction-heavy data mixing outperforms filtering for VLM training, with DCVLM-Baseline achieving 63.6% on 33 tasks for 8B models (+5.4pp over FineVision).
-
TraceAV-Bench: Benchmarking Multi-Hop Trajectory Reasoning over Long Audio-Visual Videos
TraceAV-Bench is the first benchmark for multi-hop trajectory reasoning over long audio-visual videos, showing top models reach only 51-68% accuracy with substantial room for improvement.
-
EVE: Verifiable Self-Evolution of MLLMs via Executable Visual Transformations
EVE enables verifiable self-evolution of MLLMs by using a Challenger-Solver architecture to generate dynamic executable visual transformations that produce VQA problems with absolute execution-verified ground truth.
-
MemeBench: What LVLMs Miss When Interpreting Culture-Dependent Memes
Across 26 LVLMs on MemeBench, all models show a 14.6-29.0% gap between visual coverage and cultural-knowledge coverage, and retrieval raises knowledge while lowering visual coverage.
-
TimeLens2: Generalist Video Temporal Grounding with Multimodal LLMs
TimeLens2 shows that a compact video MLLM can localize multiple evidence intervals in long videos by training on verified interval labels and a Wasserstein-based time-distance reward.
-
LongEgoRefer: A Benchmark for Long-Form Egocentric Video Referring Expression Comprehension
LongEgoRefer is a new benchmark of 1,498 referring expressions in 45-minute average egocentric videos that exposes the failure of existing Video REC models on sparse long-form spatio-temporal grounding.
-
When No Answer Is Correct: Diagnosing Absent Answer Detection for MLLMs in Video Understanding
MLLMs fail to detect absent correct answers in video QA tasks across three evaluation settings, defaulting to distractors even with chain-of-thought prompting.
-
Benchmarking Visual State Tracking in Multimodal Video Understanding
VSTAT benchmark shows state-of-the-art MLLMs perform far below humans and only modestly above answer-prior baselines on visual state tracking, failing at visual perception despite correct textual reasoning.
-
Towards Open-World Referring Expression Comprehension: A Benchmark with Training-free Multi-task Consistency Checker
OpenRef benchmark for open-world REC with F1 and N3R metrics and training-free MCC to improve existing models in complex scenarios.
-
Visual Preference Optimization with Rubric Rewards
rDPO uses offline-built rubrics to generate on-policy preference data for DPO, raising benchmark scores in visual tasks over outcome-based filtering and style baselines.
-
Learning Self-Correction in Vision-Language Models via Rollout Augmentation
Recombining pre/post-correction segments of existing rollouts yields dense self-correction training signals, giving an 8B VLM a 1.0-point average accuracy gain over GSPO on 7 benchmarks at 0.72x rollout time.
-
VISTA-Bench: Do Vision-Language Models Really Understand Visualized Text as Well as Pure Text?
VISTA-Bench shows vision-language models degrade on visualized text in images compared to equivalent pure text, with larger gaps under increased perceptual difficulty.
-
Video-OPD: Efficient Post-Training of Multimodal Large Language Models for Temporal Video Grounding via On-Policy Distillation
Video-OPD uses on-policy distillation from a frontier teacher to turn sparse episode rewards into dense step-wise signals for more efficient post-training of MLLMs on temporal video grounding.
-
HiEviDR-Bench: A Benchmark for Hierarchical Evidence Aggregation in Deep Research
A hierarchical evidence-graph benchmark reveals that multimodal deep-research models write fluent reports while failing citation, claim, and answer grounding.
-
Beyond the Eye: Efficient Multimodal Reasoning via Self-Regulated Implicit Visual Tools
Self-regulated implicit visual tools let MLLMs match or beat external-tool and latent-reasoning methods on high-resolution perception while cutting inference latency by up to 86%.
-
CAVE: Competence-Aware Visual Boundary Evidence Alignment for Video Temporal Grounding
CAVE adds boundary evidence tokens and an attention alignment reward to RL-based video temporal grounding, improving average timestamp accuracy by about 1.9 points over TimeLens-7B.
-
RefCaptioner: Multi-Reference Image-Grounded Video Captioning
Mixed-data SFT plus Hierarchical Coverage-Discounted GRPO yields open-source SOTA multi-reference image-grounded video captions on MRVBench without hurting general captioning.
-
PathScale-R1: Cross-scale Reasoning for Pathological Image Analysis
PathScale-R1 and PathScale-VQA raise cross-scale pathology VQA accuracy to 83% by training on expert multi-magnification paths with anti-shortcut curation and scale-aware RL.
-
MVEI & EmObserver: Empowering MLLM-Oriented Visual Emotional Intelligence via Emotion Statement Judgement
ESJ statement verification, the INSETS-462k corpus, the MVEI benchmark, and the EmObserver model together provide a scalable way to evaluate and improve MLLM visual emotional intelligence.
-
Self in Space: Benchmarking Self-Awareness and Spatial Cognition in UAV Embodied Intelligence
SIS-Bench shows MLLMs lag on UAV self-awareness versus spatial tasks, and motion-aware optical-flow fusion improves perception and memory.
-
Self in Space: Benchmarking Self-Awareness and Spatial Cognition in UAV Embodied Intelligence
SIS-Bench, a 4,856-question UAV benchmark, shows video MLLMs are weaker at self-awareness than spatial cognition, and optical-flow fusion improves perception and memory.
-
NEXT: Reasoning-Driven Video Recommendation via a Vision-Language Model
NEXT, a reasoning-driven video recommendation framework with a custom-trained 8B VLM, yields +0.53% watch time and +0.51% distinct exposure in production A/B tests.
-
DataComp-VLM: Improved Open Datasets for Vision-Language Models
DataComp-VLM benchmark shows instruction-heavy data mixtures outperform caption-heavy ones for VLM training, with DCVLM-Baseline reaching 63.6% on 33 tasks using 200B tokens, +5.4pp over FineVision.
-
Q-Fold: Query-Aware Focus-Context Spatio-Temporal Folding for Long Video Understanding
Q-Fold is a query-aware spatio-temporal folding technique that constructs heterogeneous focus-context inputs from long videos to improve Video-MLLM performance under fixed visual budgets.
-
Fine-grained Fragment Retrieval in Multi-modal Long-form Dialogues
Introduces FFR task, F2RVLM and FFRS models, and MLDR dataset for retrieving coherent multi-modal dialogue fragments, reporting superior performance on single-dialogue and corpus benchmarks.
-
TRON: Targeted Rule-Verifiable Online Environments for Visual Reasoning RL
TRON supplies 520 rule-verifiable online visual reasoning environments across five ability buckets that generate unlimited training instances for RL post-training, yielding consistent gains on ten external multimodal ...
-
Detect in Any Scene: An Agentic Framework for Object Detection with Experience-Aware Reasoning
DetAS-X uses an MLLM agent to adaptively compose detection workflows from restoration modules and expert detectors, enhanced by self-evolving experience harvesting, achieving substantial F1 score gains on challenging ...
-
FoodMonitor: Benchmarking MLLMs for Explainable Compliance Analysis
FoodMonitor benchmark evaluates MLLMs on explainable kitchen compliance analysis using dual-channel annotations and a composite C_score metric, with best model at 0.36.
-
MLLMs Know When Before Speaking: Revealing and Recovering Temporal Grounding via Attention Cues
MLLMs know event timing during prefill via sparse Temporal Grounding Heads but lose it in autoregressive decoding; restricting visual context to the high-attention interval at inference time improves VTG performance o...
-
Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation
Vision-OPD transfers an MLLM's privileged regional perception to its full-image policy through on-policy token-level self-distillation, yielding competitive results on fine-grained visual benchmarks.
-
Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation
Vision-OPD uses on-policy self-distillation from crop-conditioned to full-image policies within the same MLLM to close the regional-to-global perception gap.
-
Video-Zero: Self-Evolution Video Understanding
Video-Zero is an annotation-free Questioner-Solver co-evolution framework that centers self-evolution on temporally localized evidence to improve video VLMs.
-
Hide to See: Reasoning-prefix Masking for Visual-anchored Thinking in VLM Distillation
A masking-based think-answer distillation method for VLMs that selectively hides reasoning prefixes and uses self-paced scheduling to improve visual anchoring and benchmark performance.
-
Hide to See: Reasoning-prefix Masking for Visual-anchored Thinking in VLM Distillation
A reasoning-prefix masking strategy during VLM distillation encourages students to anchor their thinking on visual evidence, yielding better multimodal reasoning than prior distillation baselines.
-
Hide to See: Reasoning-prefix Masking for Visual-anchored Thinking in VLM Distillation
A new distillation method uses token-wise salient reasoning-prefix masking and self-paced scheduling to anchor student VLM thinking on visual inputs, outperforming prior distillation approaches on multimodal reasoning...
-
Video Understanding Reward Modeling: A Robust Benchmark and Performant Reward Models
Introduces VURB benchmark and VUP-35K dataset to train discriminative and generative video reward models that achieve SOTA performance on VURB and VideoRewardBench.
-
DiffCap-Bench: A Comprehensive, Challenging, Robust Benchmark for Image Difference Captioning
DiffCap-Bench supplies a diverse IDC benchmark with ten categories and LLM judging grounded in human difference lists to evaluate MLLMs more robustly than prior lexical metrics.
-
OMIBench: Benchmarking Olympiad-Level Multi-Image Reasoning in Large Vision-Language Model
OMIBench benchmark reveals that current LVLMs achieve at most 50% on Olympiad problems requiring reasoning across multiple images.
-
SSL-R1: Self-Supervised Visual Reinforcement Post-Training for Multimodal Large Language Models
SSL-R1 reformulates visual SSL tasks into verifiable puzzles to supply rewards for RL post-training of MLLMs, yielding gains on multimodal benchmarks without external supervision.
-
VISTA-Bench: Do Vision-Language Models Really Understand Visualized Text as Well as Pure Text?
Vision-language models consistently lose accuracy when a question’s text is rendered into an image, and the loss grows with perceptual difficulty like small fonts and handwriting.
-
MiMo-Embodied: X-Embodied Foundation Model Technical Report
MiMo-Embodied is a single foundation model that achieves state-of-the-art results on 17 embodied AI benchmarks and 12 autonomous driving benchmarks through multi-stage learning, curated data, and CoT/RL fine-tuning th...
-
Beyond Reasoning Gains: Mitigating General-Capability Forgetting in Large Reasoning Models
A dynamic replay and reweighting scheduler (RECAP) preserves general capabilities during RLVR while keeping reasoning performance at least as good as reasoning-only finetuning.
-
PEER: Unified Process-Outcome Reinforcement Learning for Structured Empathetic Reasoning
PEER applies GRPO reinforcement learning with a unified process-outcome reward model to structured empathetic reasoning steps on the SER dataset, yielding gains in empathy, strategy alignment, and human-likeness.
-
Aloe-Vision: Robust Vision-Language Models for Healthcare
Releases open medical LVLMs trained on a quality-filtered multimodal dataset, introduces CareQA-Vision benchmark from exams, reports performance gains over baselines, and flags adversarial vulnerabilities.
-
Latent Visual States for Efficient Multimodal Reasoning
EVA generates adaptive Latent_slot tokens as internal visual thoughts, trained end-to-end with text tokens via D-GSPO on the EVA-230K dataset, claiming performance gains and better inference efficiency.
-
Towards Open-World Referring Expression Comprehension: A Benchmark with Training-free Multi-task Consistency Checker
OpenRef benchmarks open-world REC with multi/none-target and hard scenes, and MCC is a training-free consistency plug-in that improves existing models.
-
Foresee-to-Ground: From Predictive Temporal Perception to Evidence-Driven Reasoning for Video Temporal Grounding
F2G improves video temporal grounding accuracy by decoupling event identification from boundary measurement using predictive temporal perception to create citable evidence segments for LLM reasoning.
-
Hide to See: Reasoning-prefix Masking for Visual-anchored Thinking in VLM Distillation
Proposes reasoning-prefix masking during VLM distillation to anchor student thinking on visual evidence and improve multimodal reasoning in smaller models.
-
Advancing Multimodal Judge Models through a Capability-Oriented Benchmark and MCTS-Driven Data Generation
A capability-oriented multimodal judge benchmark and MCTS-based preference-data generation improve judge models on some benchmarks, but the paper's SOTA claim is not supported by its own numbers.
-
Video-OPD: Efficient Post-Training of Multimodal Large Language Models for Temporal Video Grounding via On-Policy Distillation
Video-OPD post-trains video LLMs for temporal grounding by distilling a GRPO-trained teacher via token-level reverse KL on on-policy trajectories, outperforming GRPO with lower cost.
-
Enhancing Trustworthy GUI Grounding via Self-Critiqued Reinforcement Learning
HyperClick trains GUI grounding models with GRPO to output clicks plus confidence scores, jointly rewarding correct clicks and Brier-calibrated confidence, and reports SOTA accuracy on six of seven benchmarks with bet...
-
VisRAG2.0: Mitigating Visual Hallucinations via Evidence-Guided Multi-Image Reasoning in Visual Retrieval-Augmented Generation
EVisRAG, an evidence-guided observe-record-reason-answer pipeline trained with reward-scoped GRPO, improves multi-image visual QA accuracy by about 19% over its backbone VLM across five benchmarks.
-
Draw-In-Mind: Rebalancing Designer-Painter Roles in Unified Multimodal Models Benefits Image Editing
Rebalancing designer-painter roles by assigning design to the understanding module via the new DIM dataset yields SOTA image editing performance with a 4.6B model.
-
AIR: Adaptive Interleaved Reasoning with Code in MLLMs
AIR applies RL with a group-constrained reward and custom data pipeline to enable adaptive code-interleaved reasoning in MLLMs, reporting 6.1 pp average benchmark gains.
-
JoyAI-Image: Awaking Spatial Intelligence in Unified Multimodal Understanding and Generation
JoyAI-Image unifies visual understanding and generation via an MLLM-MMDiT architecture with spatial training signals to reach competitive benchmark performance and stronger spatial intelligence.
-
JoyAI-Image: Awaking Spatial Intelligence in Unified Multimodal Understanding and Generation
JoyAI-Image unifies visual understanding, generation, and editing in one model and claims stronger spatial intelligence through bidirectional perception-generation loops.
-
Reinforcement Learning from Human Feedback
The book introduces the origins, mathematical setup, and optimization stages of RLHF including reward modeling, reinforcement learning, rejection sampling, and direct alignment algorithms.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.