AINews Portal

Research Command Center

One ranked queue for papers, model launches, repos, agent installs, and cross-market opportunities. JSON is available at /api/research.

Opportunity

12

Paper

20

Model

20

Repo

20

Agent

19

Paper
58/1002 mentions · ▲ 10
AI Reviews Are Biased Towards the Appendix

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

Open signal
1
Paper
58/100Research papers2 mentions · ▲ 107d
AI Reviews Are Biased Towards the Appendix

Large language models used for peer review draw about half or more of their identified weaknesses from paper appendices, whereas human reviewers draw only 8.4% and focus on presentation and core ideas.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2611.1781148h mentions: 27d mentions: 2HF upvotes: 10Implementation: not detected
2
Paper
47/100Research papers2 mentions · ▲ 0
Value Over Language Model: Detecting Original Contribution in Writing

Trending research paper with fresh community or cross-source attention.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2609.0070048h mentions: 27d mentions: 2HF upvotes: 0Implementation: not detected
3
Paper
46/100Research papers0 mentions · ▲ 4182mo
LimiX-2: A Contextual Mechanism Network Towards General Structured-Data Intelligence

Trending research paper with fresh community or cross-source attention.

Inspect the implementation repo and decide whether this should become a tool/model explainer.

arXiv: 2609.1748848h mentions: 07d mentions: 0HF upvotes: 418Implementation: limix-ldm-ai/LimiX
4
Paper
45/100Research papers0 mentions · ▲ 4522mo
Scaling Automatic Research Agents via World Models

World Model RL replaces costly environment execution with a learned world model and applies debiasing and denoising to accelerate post-training of autonomous research agents.

Inspect the implementation repo and decide whether this should become a tool/model explainer.

arXiv: 2608.1256448h mentions: 07d mentions: 0HF upvotes: 452Implementation: xiyuanyang45/WMRL
5
Paper
42/100Research papers0 mentions · ▲ 38625d
The Trinity:Capability-Driven Recursive Self-Improvement for Multimodal Agents

The Trinity framework recursively improves multimodal agents by organizing tasks into a capability tree, generating simulated training worlds, and routing specialized harness modules to boost performance on diverse benchmarks.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2610.3607748h mentions: 07d mentions: 0HF upvotes: 386Implementation: not detected
6
Paper
42/100Research papers0 mentions · ▲ 38125d
Multiscale Policy Optimization for Long-Horizon LLM Agents

Multiscale Policy Optimization improves long-horizon agent training by aligning credit assignment and policy updates at the turn scale.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2610.3501348h mentions: 07d mentions: 0HF upvotes: 381Implementation: not detected
7
Paper
42/100Research papers0 mentions · ▲ 34426d
ChronoReplay: One Answer Interface Is Not Enough for Streaming Spatial Intelligence

ChronoReplay mitigates interface lock-in in spatial vision-language models by replaying heterogeneous spatial supervision chronologically, improving cross-task transfer and enabling efficient inference-time evidence selection.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2610.3428348h mentions: 07d mentions: 0HF upvotes: 344Implementation: not detected
8
Paper
40/100Research papers0 mentions · ▲ 3481mo
StreamEval: Evaluating Interactive Streaming Video Generation Models

StreamEval benchmarks interactive streaming video generation across five tasks, revealing that reliable request execution and sustained state consistency remain major challenges despite high visual quality.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2610.2840248h mentions: 07d mentions: 0HF upvotes: 348Implementation: not detected
9
Paper
39/100Research papers0 mentions · ▲ 6681mo
Waypoint-OPD: Transferring RL-Induced Policy Shifts via Gated On-Policy Distillation

Waypoint-OPD improves policy-shift distillation by gating endpoint shifts using directional agreement between early and late teacher policy changes, boosting reasoning performance with minimal overhead.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2610.2160148h mentions: 07d mentions: 0HF upvotes: 668Implementation: not detected
10
Paper
39/100Research papers0 mentions · ▲ 3541mo
MJev-Spatial: System One Spatial Intelligence via Native Attention Matching

MJev-Spatial is a fast multimodal decision model that unifies spatial reasoning tasks into a single-pass classification framework, improving speed and accuracy over autoregressive vision-language models.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2610.2227648h mentions: 07d mentions: 0HF upvotes: 354Implementation: not detected
11
Paper
36/100Research papers0 mentions · ▲ 4242mo
Atria Dawn: The Dawn of Agentic Superintelligence

Atria Dawn Preview is a foundation agentic language model trained through verified tool interactions that achieves strong benchmark results and demonstrates a shift toward human-AI project-level collaboration in scientific research.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2609.1581848h mentions: 07d mentions: 0HF upvotes: 424Implementation: not detected
12
Paper
36/100Research papers0 mentions · ▲ 6932mo
Vidu S2: Real-Time Interactive, Editable, and Spatial Video Generation

Vidu S2 introduces real-time interactive avatar and video editing models that support high-resolution spatial video generation and dynamic reference updates.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2609.1163848h mentions: 07d mentions: 0HF upvotes: 693Implementation: not detected
13
Paper
36/100Research papers0 mentions · ▲ 3772mo
NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness

NeoHorse-1 uses agentic post-training with intelligent routing, structured feedback loops, and curriculum-based distillation to improve model capabilities across agent benchmarks.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2609.0818348h mentions: 07d mentions: 0HF upvotes: 377Implementation: not detected
14
Paper
36/100Research papers0 mentions · ▲ 3772mo
Compile by Training: Turning Natural-Language Specifications into Local Neural Functions

Compile by training converts natural-language specifications into reusable neural functions by distilling teacher-generated examples into small adapters, enabling efficient deployment without remote model dependencies.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2609.0419948h mentions: 07d mentions: 0HF upvotes: 377Implementation: not detected
15
Paper
35/100Research papers0 mentions · ▲ 5122mo
Repo-To-Skill: Distilling GitHub Repositories Into AI4AI Skills

DisCo is a research agent that distills operational knowledge into reusable skills, significantly improving autonomous ML research performance across benchmarks.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2609.0274948h mentions: 07d mentions: 0HF upvotes: 512Implementation: not detected
16
Paper
35/100Research papers0 mentions · ▲ 4742mo
StudentSim: Training LLM-based Student Simulators

StudentSim trains personalized student simulators from sparse data to mirror learner responses and adapt to tutor guidance, outperforming existing models across chess, writing, and math.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2609.0159148h mentions: 07d mentions: 0HF upvotes: 474Implementation: not detected
17
Paper
35/100Research papers0 mentions · ▲ 3602mo
Qwen-Drive-1.0: An Initial Step towards a Vision-Language Foundation Model for Autonomous Driving

Qwen-Drive-1.0 is a vision-language foundation model for autonomous driving that unifies 3D perception, visual question answering, and motion planning via shared representations and staged training.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2609.0011148h mentions: 07d mentions: 0HF upvotes: 360Implementation: not detected
18
Paper
35/100Research papers0 mentions · ▲ 3743mo
StateM: Reaching 95.3% Raw Accuracy, or a \$15 Frontier Run, on Terminal-Bench 2.1 via Harness Scaling

StateM is a runtime system that improves long-horizon agent execution through durable states, recoverable runbooks, and enforceable procedural controls without altering model weights.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2608.1508948h mentions: 07d mentions: 0HF upvotes: 374Implementation: not detected
19
Paper
35/100Research papers0 mentions · ▲ 5533mo
BDH-CQ: In-Context Learning with Recurrent Latent Reasoning

A 150M-parameter reasoning model using recurrent latent reasoning and in-context learning achieves a new cost-accuracy frontier on ARC-AGI-1.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2608.0988848h mentions: 07d mentions: 0HF upvotes: 553Implementation: not detected
20
Paper
35/100Research papers0 mentions · ▲ 3363mo
Kimi K3: Open Frontier Intelligence

Kimi K3 is a 2.8-trillion-parameter open-source AI model that uses a Mixture-of-Experts design to activate only 104 billion parameters per token, making it efficient despite its size. It introduces new attention mechanisms and a stable expert-routing method to improve performance across long sequences and deep networks, achieving roughly 2.5x better scaling efficiency than its predecessor. The model excels at long-horizon coding, agentic tasks, reasoning, and vision, outperforming all other open models and most proprietary ones, though it still trails the top closed models like Claude Fable 5 and GPT-5.6 Sol. Its release as open weights aims to accelerate research and deployment of frontier-level AI.

Read the abstract and check whether the method changes tool, benchmark, or model coverage.

arXiv: 2607.2465348h mentions: 07d mentions: 0HF upvotes: 336Implementation: not detected

Scores are synthesized from the portal's own tables: Radar, papers, model releases, repo snapshots, and MCP/skill discovery. Use this as an editorial triage queue, then verify primary sources before publishing.