Getting more from several models than from one

Claims

• observationsingle paper↑1↓0
  In a 5-agent LLM debate where agents iteratively revise a numeric preference vector, switching from all-to-all to ring communication once aggregate disagreement falls below a threshold reaches consensus using fewer tokens than all-to-all throughout and in fewer rounds than ring-only, but it does not dominate for every backbone: with Gemini 2.5-Flash all-to-all throughout converged in fewer rounds than the adaptive policy.
• observationsingle paper↑1↓0
  Across 41 models and seven multi-agent benchmarks, gains in single-agent capability translate only weakly into multi-agent competence (understanding other agents, joint planning, efficient communication, adaptation); on the multi-agent planning task accuracy rose only about 30% across a family's full scale range, so coordination ability does not emerge from scaling alone and needs its own data and training.
• mechanismmechanism reasoning↑1↓0
  Adding more models to an LLM ensemble yields diminishing marginal gains and past some ensemble size the marginal gain turns negative, so aggregate performance degrades with further additions.
• observationsingle paper↑1↓0
  For CAE simulation tasks (OpenFOAM, FEniCS, PyChrono, COMSOL) run with a current general-purpose coding-agent harness and a strong backbone, a single agent with no multi-agent role decomposition and no scripted self-critique matches or exceeds the published scores of multi-agent specialized CAE systems, because the harness itself supplies the multi-turn reasoning and execution-feedback repair the decomposition was built to compensate for.
• observationsingle paper↑1↓0
  On complex clinical diagnosis from published case reports (rare disease and challenging cases), running several presenter-rebutter debater pairs on one backbone and synthesizing their revised hypotheses with an aggregator raises exact-match diagnostic accuracy and rated safety over the same model prompted once, while giving no measurable gain on multiple-choice clinical knowledge questions.
• mechanismsingle paper↑1↓0
  For deepfake video detection with 7-8B open multimodal backbones, splitting the analysis into four separately prompted perspective agents (texture, lighting, motion, physics) whose reports a judge agent reconciles beats the same backbone doing chain-of-thought or multi-turn self-questioning over the same four perspectives, and the gain is not explained by extra turns or repeated frame access.
• observationsingle paper↑1↓0
  In embodied multi-agent household planning with a fixed high-level skill interface, a team's task success rate overstates coordination quality, because goal predicates can be reached through illegal shortcuts: several of 11 tested MLLMs succeed on more episodes than they produce legal plans on (Claude Haiku 4.5 63.1% success vs 47.2% legal; Qwen3.6-Plus 69.1% vs 56.7%), while others honour constraints yet miss the goal.
• observationsingle paper↑1↓0
  On four diagram-grounded geometry benchmarks a multi-agent pipeline consistently helps open-source vision-language models (Qwen-2.5-VL 7B +6.8 and 32B +3.3 points on Geometry3K, with further gains on OlympiadBench and We-Math) but a strong closed model (Gemini 2.0 Flash) generally did better as a single agent on the classic benchmarks, gaining only modestly on the newest one, so agentic decomposition pays for weaker models and less familiar tasks rather than universally.
• observationsingle paper↑1↓0
  In fully connected networks of LLM agents (Llama 3.2-8B, 50 agents, 10 rounds) where a fixed minority holds an unchanging extreme stance, neutral agents' textual justifications converge on the biased agents' vocabulary faster and more completely than their numeric stances do, so monitoring only the text overstates how much the group actually agrees.
• observationsingle paper↑1↓0
  In a generational genetic search over LLM-agent-generated scientific hypotheses (drug repurposing across 34 cancer types), replacing fitness-guided parent selection with uniform random selection while holding the scientific roles, variation operators and hypothesis count fixed lowers the population's mean and minimum external evidence scores but leaves the maximum unchanged, and removing crossover or mutation altogether produces mixed shifts that do not survive multiple-testing correction — so the selection rule, not the semantic variation operators, is where the measurable coordination gain sits.
• mechanismsingle paper↑1↓0
  In a hierarchical multi-agent system where a controller gates acceptance of worker outputs on semantic entropy and semantic density computed over resampled completions, the intervention helps only where failures come from semantic instability (vague or contradictory outputs) and hurts where they come from evidence selection or reasoning-chain errors \u2014 gaining on ambiguity/hallucination-prone benchmarks while losing 10-14 points on multi-hop QA.
• observationsingle paper↑1↓0
  When initially identical LLM agents act in a shared persistent world where their artifacts remain and are scored by an external simulator, the shared society builds broader and more resilient technology portfolios than a matched best-of-N population of isolated agents, but isolated search still yields the single best-performing artifact.
• observationsingle paper↑1↓0
  On knowledge, math and medical QA benchmarks with 8B open backbones (Llama-3.1-8B-Instruct, Qwen3-8B), the accuracy advantage of multi-agent debate pipelines over a single agent can be matched by a single-turn single agent whose hidden state is steered by a query-composed mixture of role vectors, at roughly a twentieth of the token cost, so the benefit of these pipelines comes from multi-perspective specialization rather than from inter-agent text exchange.
• mechanismsingle paper↑1↓0
  In label-free RL post-training of reasoning models, deriving each model's reward from a separately-trained peer model's majority vote — rather than from its own majority vote — avoids the reward collapse and length degeneration that self-rewarding runs show, and the benefit grows as the paired models differ in family or size because their errors overlap less.
• observationsingle paper↑1↓0
  For an LLM acting as orchestrator of dynamically created subagents, summarizing its own past trajectories into reusable workflow skills, short operational tricks, and capability cards for the candidate subagent backbones, then injecting those into the orchestrator's context, raises task scores on delegation- and aggregation-heavy tasks, and the subagent model cards are the component whose removal costs most.
• mechanismsingle paper↑1↓0
  For LLM agents classifying nodes in text-attributed graphs, serializing sampled neighborhoods into prompt text makes the prediction depend on the arbitrary order of neighbors; replacing that text with a fixed-size, permutation-invariant vector of graph statistics (degree, PageRank, clustering, cross-community fraction, label entropy, majority-label fraction) plus only the top-k semantically relevant node texts raises accuracy and cuts per-message tokens.
• observationsingle paper↑1↓0
  For LLM multi-agent systems failing on long-horizon web tasks, re-running the whole task from scratch — plain, with self-reflection, or with a critic agent — repairs few failures (at most 6.90% within three attempts, with feedback variants below plain rerun), while reconstructing the recorded execution prefix and intervening once at a symptom-flagged trace node repairs 20.15% with a single attempt.
• mechanismsingle paper↑1↓0
  In LLM multi-agent systems that reflect after failure, restricting reflection to the single agent responsible for the first incorrect step — located by a trained step-level correctness scorer and paired with a counterfactually generated corrected step — raises task success more than having every agent reflect on the trajectory, because reflections written by agents that behaved correctly enter their memory as misleading guidance for later tasks.
• observationsingle paper↑1↓0
  In LLM multi-agent systems whose communication graph is produced by a learned topology generator, the generated graph carries edges that are not needed: scoring each edge by how much masking it changes the task outcome and the semantic entropy of the final answer, then keeping a top-ranked budgeted subgraph, cuts online tokens by about 20-25% while leaving average accuracy equal or slightly higher, and a small distilled predictor of those subgraphs transfers to generators and agent counts it was not calibrated on.
• mechanismsingle paper↑1↓0
  In multi-agent debate with 3 open-weight agents over 3 rounds, giving each agent a memory of past debates helps only if retrieval is conditioned on the current consensus ratio and peer responses are annotated with memory-derived reliability labels; randomly inserting past cases is the weakest retrieval policy and plain task-similarity retrieval trails every consensus-conditioned variant.
• observationsingle paper↑1↓0
  In multi-agent literature-synthesis pipelines, most of the gain over a fixed role-based pipeline comes from the shared-workspace architecture and the added critique/dissemination actions rather than from learning the coordination policy: on OARelatedWork, a learned decentralized RL controller scored 3.82 overall versus 3.80 for a centralized LLM coordinator given the same workspace and action space, while a naive orchestrator without that architecture scored 3.40.
• mechanismsingle paper↑1↓0
  In multi-agent LLM networks where a correct minority peer disagrees with an agreeing majority, ranking peers by raw divergence from the receiver's own belief systematically prefers the wrong peers, while scoring each peer by asserted precision divided by its divergence from a second-order model of what that peer should have concluded given its data subscription isolates the correct one.
• mechanismsingle paper↑1↓0
  For multi-agent LLM pipelines on multimodal medical referral from real inpatient records, requiring agents to exchange typed logical edges (causal cascade, mutual reinforcement, exclusion, conditional context) in a shared reasoning graph instead of free-form dialogue raises referral F1, and the causal-chain edge type carries most of that gain.
• observationsingle paper↑1↓0
  In multi-agent LLM pipelines over reasoning and code tasks with a pool of 3B-32B open-weight backbones, conditioning each per-step routing decision on a learned filtered memory rather than the full accumulated trajectory leaves accuracy about the same while cutting inference cost by roughly 40-50%, so most of the accumulated history is cost without accuracy value.
• observationsingle paper↑1↓0
  In multi-agent LLM workflows with a mixed-size model zoo, choosing which model instantiates each agent step-by-step from an online estimate of task-progress gain plus remaining time and cost budget keeps task quality near the best single-model policy at lower energy than one-shot query-level routing or stall-triggered escalation.
• mechanismsingle paper↑1↓0
  For multi-agent pipelines on knowledge-intensive QA with an 8B backbone, deciding the collaboration topology after retrieval and conditioning it on the retrieved evidence yields higher accuracy than generating the topology from query semantics alone with the same retrieved evidence available at execution.
• observationsingle paper↑1↓0
  When multi-agent pruning methods (AgentPrune, AgentDropout, AGP, DAGP, GPTSwarm) are re-run under a fixed backbone, agent registry, role-to-node assignment and runtime, their accuracy-cost gains do not hold across starting topologies and mostly do not beat random pruning at the same structural budget — averaged over 20 cells the optimized systems lost 0.77 accuracy points against the original system versus 0.44 for matched random pruning, and only 12 of 25 method-topology pairs beat their random control.
• observationsingle paper↑1↓0
  On multimodal cultural-interpretation multiple-choice questions, pipelines that pool several same-backbone evidence streams (perception, retrieval, web verification) and weight them by scored relevance beat majority voting, self-consistency and agent debate mainly on the items where the streams disagree, and the margin shrinks sharply as the backbone gets stronger.
• observationsingle paper↑1↓0
  When one agent's context carries planted misinformation, the wrong answer persists through multi-agent debate and other agents often adopt it, yet debate still degrades less than a single misinformed agent (relevant misinformation cut single-agent accuracy by 16-25% on CWQ), especially when most agents are unexposed; consensus protocols were more stable than voting under peer pressure, and a misinformed majority can steer the group.
• mechanismsingle paper↑1↓0
  In one-shot, no-communication self-play where independent copies of the same LLM are told their counterparts run the same model, models exceed the Nash baseline mainly in games whose optimal joint play requires matching actions (coordination, stag hunt, public goods) and fall below it in games requiring symmetry-breaking (anti-coordination, battle of the sexes), with several open-weight models scoring near the worst joint cell.
• observationsingle paper↑1↓0
  When RL-training a central-agent multi-agent harness with 1.5B-3B Qwen2.5 backbones on multi-hop QA, optimizing only the coordinating central policy and leaving sub-agents frozen scores higher than jointly training central and sub-agent policies (average F1 47.79 vs 45.36, accuracy 42.88 vs 40.49 at 3B).
• observationsingle paper↑1↓0
  When routing each image to one of several domain-specialist vision backbones trained on a shared label space, a prompted MLLM router (Gemma-4-12B) reaches nearly the accuracy of a trained neural-network router and beats majority-vote fusion, because routing only needs coarse domain identification, which a general multimodal model already does zero-shot.
• observationsingle paper↑1↓0
  For routing a video-production request to an ordered chain of registered sub-agents, an 8B open model fine-tuned on descriptor-conditioned routing plans (SFT on verified plans plus GRPO rewarding validity, sub-agent coverage F1 and dependency-consistent order) picks chains more accurately than much larger prompted planners, with the gap widest on long, structurally unfamiliar chains.
• observationsingle paper↑1↓0
  In simulated networks of LLM agents where one or two agents are instructed to advocate a position, the messages other agents write understate how far their elicited stance has moved: they mirror the persuader's rhetorical framing in most messages while explicit stance-change language is rare, so monitoring only generated text misses belief movement that a repeated token-probability probe detects.
• mechanismsingle paper↑1↓0
  Splitting a task across several agents helps on long-horizon tasks whose parts depend on each other sparsely, and hurts on tightly coupled sequential workflows, where generic multi-agent methods fall below a single-agent harness because decomposition fragments the dependency chain and coordination overhead outweighs the context isolation it buys; as single agents improve the collaboration benefit shrinks.
• mechanismsingle paper↑1↓0
  For teams of up to 50 heterogeneous embodied LLM agents in simulated wildfire missions, building the manager hierarchy per mission so that concurrently executable work sits under a parallel-allocating manager and prerequisite-chained work under a phase-advancing manager beats fixed decentralized, single-planner, and hybrid coordination architectures on mission score and on progress accumulated per step, and the advantage does not depend on which base language model is used.
• observationsingle paper↑1↓0
  For translating heterogeneous MOEA implementations from MATLAB/PlatEMO into tensorized GPU code, a multi-agent pipeline that reconstructs source semantics into a shared blueprint, generates several strategy-biased candidates in parallel, and repairs them from execution feedback converges more often than one-shot translation with the same LLM backend, on all three backends tested.
• observationsingle paper↑1↓0
  In a two-agent extract-then-answer pipeline on multilingual extractive QA, routing the inter-agent handoff through English (forward-translate facts, answer in English, back-translate) lowers Exact Match relative to keeping handoffs in the user's language, and the loss grows with distance from English (13.0 points Spanish up to 30.6 points Hindi with Aya-23-8B).

Techniques

• Arbitrate conflicting evidence streams by relevance-gated fuprocess
  Generate perception, retrieval and web-verification evidence with the same backbone, score each stream for relevance/accuracy/discriminativeness, softmax-weight them into a fused answer, and refine under six cultural rubric criteria.
• Audit agent influence with belief probes and exposure logsprocess
  Instead of reading agent messages, score each agent's stance after every round with a fixed token-probability probe and log who saw which content from whom, so influence can be attributed to direct or peer-relayed exposure.
• Compare agent pruning against random pruning at equal budgetprocess
  For every structural change a multi-agent optimizer emits, build a random topology with the same node/edge budget and evaluate it under identical model, agents, roles and runtime, so the credit goes to the optimizer rather than to having de
• Compose role steering vectors per query instead of debatingtraining
  A learnable codebook of role steering vectors, warm-started from contrastive activation directions and fused by a query-aware top-K router, is added to the last prompt token's residual stream at the middle layer of a frozen LLM, trained by
• Coordinate agents through a persistent shared world, not mesprocess
  Let role-free agents build durable, executable artifacts in a shared simulated environment so later agents inherit and modify them, instead of coordinating through conversation or assigned roles.
• Coordinate hypothesis agents with a generational genetic algprocess
  Specialized LLM agents generate, pairwise-score and semantically recombine or mutate structured hypotheses, with a genetic algorithm doing tournament parent selection and joint parent-offspring truncation to a fixed population.
• Debate-Mixture-of-Agents: rebut, revise, then aggregateprocess
  Each debater runs a presenter that drafts a diagnosis, a rebutter that attacks it for missing evidence and unsafe omissions, and a presenter revision; a separate aggregator then merges all revised hypotheses without forcing consensus.
• Drop CAE scaffolding, supply solver tutorials insteadprocess
  Run CAE simulation tasks with an off-the-shelf coding-agent harness (file, shell, multi-turn, execution feedback) and no simulation-specific roles, retrieval pipeline or reflection prompt, while requiring the agent to read the relevant solv
• Exchange typed logical edges, not dialogue, between agentsprocess
  Each modality agent builds a local evidence graph and agents update one another only by establishing typed edges (cascade, syndrome, exclusion, context) before an expert agent arbitrates the merged candidate graphs against a referral knowle
• Gate worker outputs on semantic entropy plus densityprocess
  The orchestrator resamples each worker response, computes semantic entropy and embedding-space semantic density, and accepts only if both clear thresholds, otherwise reprompts, reassigns, or delegates in parallel.
• Generate the agent topology after retrieval, conditioned on process
  Retrieve evidence first, then autoregressively generate the multi-agent DAG (roles and edges) conditioned on the retrieved evidence and its provenance, with a generator trained on execution-scored, cost-pruned candidate topologies.
• Keep inter-agent handoffs in the user's languageprocess
  Have agents pass extracted facts and answers to each other in the user's language instead of pivoting through English translation.
• Learn agent action selection instead of scripting a pipelineprocess
  Each LLM agent reads a 20-dimensional summary of its private notes, shared knowledge, draft and feedback and picks Retrieve, Disseminate, Compose or Critique under a shared IPPO policy, instead of being assigned a fixed pipeline role.
• Match the agent hierarchy to task interdependenceprocess
  Construct a per-mission manager tree in which horizontal managers hand out concurrent assignments to their children and vertical managers decompose the mission into ordered phases and advance only when completion conditions are met, with re
• Pass graph structure as a permutation-invariant signature, nprompting
  Summarize sampled graph neighborhoods for an LLM agent as a fixed-length vector of named graph statistics, keeping raw text only for the top-k semantically relevant nodes, so the observation is order-independent and does not grow with neigh
• Rank communication edges by masking effect, then prune to buprocess
  Mask each communication edge in a generated multi-agent graph, score it by the drop in task score plus the rise in semantic entropy of the final response, keep the top-ranked edges under a pruning budget, and distill those subgraphs into a
• Reflect only the agent that made the first wrong stepprocess
  A process-reward model scores each step of a failed multi-agent trajectory, the first low-scoring step and its agent are identified, a corrected version of that step is generated and verified, and only that agent writes a reflection into it
• Replay extracted skills, tricks and model cards to the orcheprompting
  Run the swarm, summarize trajectories into SKILL.md workflow patterns, a trick library, and per-backbone model cards, and reinject all three into the orchestrator's prompt at inference.
• Replay the recorded prefix, then intervene at the symptomatiprocess
  Log every model and tool boundary interaction as an event-dependency graph, replay it deterministically up to a node flagged by symptom rules plus a semantic judge, inject a repair instruction conditioned on the symptom category, and resume
• Retrieve debate memory by consensus level, then reweight peeprompting
  Each debating agent retrieves past-debate cases with an MMR trade-off whose relevance/diversity coefficient is set by the current consensus ratio (high consensus pulls in contrastive cases), and labels each peer's response high- or low-conf
• Reward each model by a peer model's majority vote, not its otraining
  Train several parameter-disjoint models simultaneously with RL, where each model's rollouts are scored against the majority-vote answer of a different model in the cohort, passed along a directed ring.
• Route each agent step by predicted task-progress gainprocess
  Scores intermediate workflow state into a progress value, predicts each candidate model's progress gain from structured and semantic features, and picks the model maximizing gain minus time/cost-budget penalties at every dispatch step.
• Route images to expert backbones with a prompted MLLMprocess
  An MLLM is prompted with the image plus blur/brightness/contrast/noise statistics and outputs one of five domain labels, dispatching the image to the matching fine-tuned vision expert, with an UNSURE option falling back to the best balanced
• Route multi-agent steps over a gated execution memoryprocess
  A jointly trained router picks each step's role and backbone from a memory kept compact by learned write and retrieval gates, and halts when the memory holds enough evidence.
• Score coordination constructs separately from task successprocess
  Alongside the binary goal check, compute per-trajectory construct scores from simulator-logged events — makespan versus an oracle reference, precedence-violation rate with a legality gate, shared-resource collision rate, and buffer overflow
• Shared semantic blueprint, parallel restructuring branches, process
  One agent reconstructs the source program's states, operators and dependencies into a structured representation and a target-side blueprint; several agents then generate alternative tensor realizations under that blueprint, failures are rep
• Size an ensemble by its marginal-gain (benefit yield) curveprocess
  Treat ensemble performance as a function of member count and stop adding models at the size where marginal gain per added model crosses zero.
• Split forensic analysis into per-cue agents, reconcile with process
  Four MLLM agents each inspect the same sampled video frames for one forgery cue type and report observations without a verdict; a judge agent weighs the four reports (optionally plus the frames) and issues the real/fake decision and explana
• Switch from dense to sparse agent topology as disagreement fprocess
  The harness selects all-to-all inter-agent communication while aggregate disagreement exceeds a threshold and a ring topology once it falls below, with a token budget decremented each round and a guard that halts on exhaustion.
• Train only the central coordinating agent, freeze sub-agentstraining
  Treats sub-agents as part of the environment and applies RL gradients only to the central agent's policy, with trajectories reconstructed from captured per-session interface call records.
• Train the orchestrator on descriptor-conditioned routing platraining
  Fine-tune the planning model to emit a rationale plus an ordered sub-agent plan from tool descriptors, then refine with GRPO rewarding parseable output, coverage F1 against the target plan, and satisfied precedence edges.
• Weight peers by cognitive SNR, not by divergenceprocess
  Each incoming agent message is weighted by the sender's asserted belief precision divided by the KL divergence between its reported belief and what the receiver's second-order model predicts a peer with that telemetry access should have rep

Sources

Every paper, post or observation behind this capability. Ones cited by a claim show the claim and its stance; ones filed directly are about the capability but have no claim drafted from them yet. Open one to add support or contest it.

• Rethinking Multi-Agent Collaboration: When More Is Less2026-09-17
  ↑ supports a claim here
• Bias Amplification in Multi-Agent Network: How Biased Agents Shape Opinions and Rhetoric2026-09-16
  ↑ supports a claim here
• The Universe of Universes: Benefit Yield Functions, Implosion Thresholds, and Infrastructure-Aware Optimization in Multi-LLM Systems2026-09-14
  ↑ supports a claim here
• HypoEvolve: Genetic Algorithms Enable Multi-Agent LLMs to Discover Scientific Hypotheses2026-09-14
  ↑ supports a claim here
• Assembling the CREW: A Collaborative Multi-agent Reinforcement Learning Framework for Automated Related Work Generation2026-09-14
  ↑ supports a claim here
• Translating the Translator: Decomposing the Cost of English-Forced Inter-Agent Communication2026-09-14
  ↑ supports a claim here
• EMR: Self-Evolving Medical Multi-Agent System via Experience Mining and Reuse2026-09-14
  about this capability · no claim drafted yet
• ORCH: Organizational Principles Enable Collective Intelligence in Embodied AI2026-09-10
  ↑ supports a claim here
• Multi-Agent Agentic Graph Learning via Structural Signatures2026-09-09
  ↑ supports a claim here
• Rethinking the Evaluation of Efficiency Methods for Multi-Agent Systems2026-09-05
  ↑ supports a claim here
• A Structured Debate-Mixture-of-Agents Framework for Complex Clinical Diagnostic Decision Support2026-09-04
  ↑ supports a claim here
• What Do CAE Simulation Agents Really Need Beyond a Generic Harness?2026-09-03
  ↑ supports a claim here
• Remember and Reweight: Enhancing Multi-Agent Debate with Experience Memory and Confidence Estimation2026-09-03
  ↑ supports a claim here
• Semantics-Guided Automatic Tensorization for Multiobjective Evolutionary Algorithms: A Multi-Agent Framework2026-09-02
  ↑ supports a claim here
• Agents That Model Agents: Five Principles Toward a Theory of Mind for 6G Networks2026-09-01
  ↑ supports a claim here
• SwarmBench: Can Large Language Models Act as Agent Swarm Orchestrators?2026-08-31
  ↑ supports a claim here
• Evidence, Logic, and Compliance: Multi-Agent Structured Graph Reasoning with Expert Arbitration for Medical Referral2026-08-31
  ↑ supports a claim here
• Learning What to Retain: Gated-Memory Routing for Efficient Collaboration in Multi-Agent LLM Systems2026-08-31
  ↑ supports a claim here
• CM2: Multimodal Cultural Reasoning via an Integrated Multi-Agent Framework2026-08-31
  ↑ supports a claim here
• HiRS-Agent: A Hierarchical Multi-Agent System for Reliable Long-Horizon Remote Sensing Task Solving2026-08-31
  about this capability · no claim drafted yet
• Harness-RL: Black-Box Reinforcement Learning with Action-Args Decoupling for Central-Agent Multi-Agent Harnesses2026-08-30
  ↑ supports a claim here
• FRAMEWORKERS: A Dynamic Multi-Agent Framework for AI-Generated Video Production2026-08-30
  ↑ supports a claim here
• CoCoBench: A Cooperative Coordination Benchmark for Embodied Multi-Agent Task Planning2026-08-28
  ↑ supports a claim here
• Finding Where the Buck Stops: An Automated Failure Attribution-Based Reflection Framework for Multi-Agent Collaboration2026-08-28
  ↑ supports a claim here
• When Evidence Shapes Collaboration: Knowledge-Conditioned Topology Generation for Multi-Agent Systems2026-08-28
  ↑ supports a claim here
• One Model, Many Minds: Unlocking Multi-Agent Synergy in a Single Agent via Mixture of Roles2026-08-27
  ↑ supports a claim here
• SwarmWorld: Stigmergic technological evolution in societies of language-model agents2026-08-26
  ↑ supports a claim here
• Repair or Resample? Rethinking Failure Debugging in LLM Multi-Agent Systems2026-08-26
  ↑ supports a claim here
• ProgRouter: Online Progress-Guided Orchestration for Multi-Agent LLM Workflows under Quality-Cost Tradeoffs2026-08-26
  ↑ supports a claim here
• A Control-Theoretic Approach for Resource-Aware Consensus in Multi-Agent AI2026-08-25
  ↑ supports a claim here
• Belief Cascades Drive Persuasion in LLM Agent Networks2026-08-25
  ↑ supports a claim here
• Co-RL: Unsupervised Reasoning Emerges from Diverse Cohort in Multi-agent RL2026-08-18
  ↑ supports a claim here
• Discovering Efficient and Explainable Communication Topologies for LLM-based Multi-Agent Systems via Causal Inference2026-08-13
  ↑ supports a claim here
• MLLM-Routed Heterogeneous Ensembles for Robust Cross-Dataset Image Classification2026-08-13
  ↑ supports a claim here
• Do LLMs Beat Nash? Testing Decentralized Coordination in Self-Play Multi-Agent Games2026-08-12
  ↑ supports a claim here
• Advancing MLLM-based UAV Image Understanding and Reasoning: A Benchmark and a Training-Free Multi-Agent System2026-08-12
  about this capability · no claim drafted yet
• Semantic Uncertainty-Guided Orchestration in Hierarchical Multi-Agent Systems2026-08-11
  ↑ supports a claim here
• Multi-Agent Forensic Reasoning for Generalizable Deepfake Video Detection2026-08-07
  ↑ supports a claim here
• Contextual Value Alignment via Multilayer Combinatorial Fusion2026-08-07
  about this capability · no claim drafted yet
• Misinformation Propagation in Benign Multi-Agent Systems2026-06-15
  ↑ supports a claim here
• Do Multi-Agents Solve Better Than Single? Evaluating Agentic Frameworks for Diagram-Grounded Geometry Problem Solving and Reasoning2025-12-18
  ↑ supports a claim here
• Single-Agent Scaling Fails Multi-Agent Intelligence: Towards Foundation Models with Native Multi-Agent Intelligence2025-12-09
  ↑ supports a claim here