DynaFlow: Learning Free-Form Multi-Agent Orchestration

Ruhao Liu; Qi Li; Shizun Wang; Xinchao Wang

cs.MAOct 25, 2026

DynaFlow: Learning Free-Form Multi-Agent Orchestration

Authors: Ruhao Liu, Qi Li, Shizun Wang, Xinchao Wang

Organizations: National University of Singapore

Abstract

Multi-agent systems often rely on predefined collaboration structures, limiting their ability to adapt to diverse task requirements. We introduce DynaFlow, a framework that integrates free-form executable workflow synthesis into the action space of the main agent. The agent decides when collaboration is needed and generates task-specific programs that coordinate sub-agents through sequential or parallel execution, structured information exchange, and verification. Execution results and traces inform subsequent actions, enabling the agent to adapt collaboration as reasoning progresses. We train only the main agent through supervised fine-tuning on distilled trajectories, followed by refinement on examples filtered for unsupported premises, biased verification, and information-flow defects. All sub-agents remain fixed, separating orchestration learning from executor capability improvement. Across six benchmarks spanning mathematical reasoning, scientific question answering, and multi-hop retrieval, DynaFlow achieves 88.08% overall accuracy in Forced mode, outperforming the strongest evaluated baseline by 10.90 percentage points with the same Qwen3.5-27B backbone. Workflow analysis finds more explicit use, verification, and refinement of upstream outputs after training, highlighting cross-stage information flow as a key dimension of multi-agent orchestration.

Figures & tables

Figure 1: Overall accuracy versus collaboration size. Accuracy versus the mean number of agent roles per question on 688 questions (Qwen3.5-27B). In Forced mode, DynaFlow improves accuracy from Base’s 75.9% to 88.1%. Dynamic achieves 87.9% with 3.44 agent roles per question on average, while answering 47% of questions without workflows.

Figure 2: Overview of DynaFlow. (1) Two-stage SFT trains only the Main Agent on first-turn reasoning

r

and workflow action

a_{1}

, using distilled demonstrations followed by a curated subset. (2) It chooses to answer directly, call a tool, or generate a JavaScript workflow. (3) Frozen sub-agents collaborate within the workflow according to the Main Agent’s program. (4) Outputs and execution traces update the history for subsequent decisions.

Single-agent baselines
Method	AIME24	AIME25	AIME26	PolyMATH	GPQA-D	HotpotQA	Overall
CoT	96.67	93.33	86.67	83.00	85.35	46.50	74.27
CoT-SC	90.00	93.33	96.67	84.50	85.86	50.50	76.16
Self-Refine	96.67	93.33	93.33	87.50	88.38	48.00	77.18
Automated multi-agent baseline
AFlow	93.33	96.67	100.00	86.00	87.37	47.00	76.45

Table 1: Main results. Accuracy (%) on six benchmarks. All systems use Qwen3.5-27B, and all workflow settings share the same frozen sub-agents and runtime. Overall is micro-averaged over all 688 questions. Runs that end without a valid final answer count as incorrect. Bold and underline mark the best and second-best result in each column. with the single-agent baselines (§ 5.1 ).

Figure 3: Workflows help only after training. Accuracy change in Forced mode relative to Base, Direct. Without training, workflows lower accuracy on every benchmark except HotpotQA; after training, Forced mode matches or exceeds Base, Direct on all six.

Figure 4: Workflow design quality. Judge scores for the first workflow generated for each question in Forced mode across all benchmarks: 0 = major defect, 1 = minor weakness, and 2 = no defect.

Main Agent	Direct	Dynamic	Forced
Base	71.51	65.12	75.87
Stage 1	76.74	84.45	85.47
DynaFlow (Stage 2)	77.18	87.94	88.08

Table 2: Training stage

\times

execution mode. Overall accuracy (%). Only the Main Agent differs between rows. Per-benchmark results are in Table 6 .

Figure 5: Workflow structure and information use across stages. Distributions (%) of six properties of workflow programs

P

generated under Forced execution: (a) execution structure, (b) role specialization, (c) dependency pattern, (d) output reuse, (e) aggregation strategy, and (f) sub-agent count. Definitions are provided in Table 5 .

Figure 6: Task-specific role allocation. Percentages of executed sub-agent nodes by role and benchmark family in DynaFlow’s Forced mode. The three AIME years are pooled;

n

denotes question counts. Classification details are in Appendix D .

Appendix figures & tables8 assets

Supplementary material from the paper’s appendix.

Appendix

Parameter	Stage 1	Stage 2
Training examples	2,257	712
Learning rate	$5\times 10^{-5}$	$1\times 10^{-5}$
Epochs	2	2
LoRA rank / $\alpha$ / dropout	16 / 32 / 0
Optimizer	Fused AdamW
Schedule / warmup ratio	Cosine / 0.05

Appendix

Table 3: Training configuration. Only the Main Agent is updated. Shared values apply to both stages; maximum sequence length denotes the configured upper limit.

Main Agent
Parameter	Value
Context length	131,072
Max output tokens	81,920
Temperature / top- $p$ / top- $k$	1.0 / 0.95 / 20
Presence penalty / seed	1.5 / 42
Sub-agents

Appendix

Table 4: Evaluation decoding parameters.

Dimension	Categories and definitions
Execution structure	Code only : no agent reasoning node, only code or tools; Single agent : exactly one agent reasoning node; Serial : multiple agents operate mainly in sequence, with later stages depending on earlier ones; Parallel : agents operate mainly independently in parallel, without substantial sequential dependencies; Hybrid : parallel branches and sequential agent dependencies are meaningfully combined.
Role specialization	None : fewer than two agents; Homogeneous : agents mainly perform the same role or type of attempt; Complementary : agents differ in roles, perspectives, subtasks, or methods; Mixed : homogeneous attempts and complementary roles coexist.
Dependency pattern	None : no agent depends on another agent’s output; Chain : outputs mainly form a sequential chain; Fan-out : one upstream result feeds multiple downstream agents; Fan-in : one downstream agent combines multiple upstream results; Multi-stage : multiple dependency forms or sequential stages are combined.
Output reuse	None : no upstream output is passed downstream; Passed : an upstream output is supplied, but its meaningful use is not explicitly required; Consumed : the actual upstream output is supplied and the downstream instruction explicitly requires using, analyzing, comparing, refining, critiquing, or transforming it.
Aggregation strategy	None : no result aggregation; Concatenation : results are collected without substantive comparison or synthesis; Selection : one result or candidate is selected; Voting : results are combined by voting or majority agreement; Comparison : outputs are explicitly compared; Synthesis : outputs are integrated into a new answer or representation; Correction/refinement : an upstream result is reviewed and may be revised; Mixed : multiple substantive aggregation mechanisms are used.
Agent count	Number of distinct agent reasoning nodes explicitly instantiated in the program, excluding deterministic code, formatting, and simple tool calls. Counts are displayed as 0, 1, 2, 3, 4, and 5+ in Figure 5 .

Appendix

Table 5: Workflow structure definitions. Six structural dimensions used to characterize the workflow programs

P

generated by the Main Agent.

Main Agent	Mode	AIME24	AIME25	AIME26	PolyMATH	GPQA-D	HotpotQA	Overall
Base	Direct	93.33	93.33	93.33	74.00	83.33	47.50	71.51
	Dynamic	63.33	63.33	73.33	64.00	77.78	53.00	65.12
	Forced	76.67	70.00	70.00	70.00	77.78	81.50	75.87
Stage 1	Direct	93.33	96.67	90.00	87.50	87.88	47.50	76.74
	Dynamic	96.67	100.00	93.33	88.50	82.83	76.50	84.45
	Forced	96.67	96.67	100.00	84.50	84.34	82.00	85.47

Appendix

Table 6: Results by training stage and execution mode. Per-benchmark accuracy (%) by Main Agent and execution mode. DynaFlow is Stage 1 followed by curated training (Stage 2). Shaded rows denote DynaFlow.

Figure 8: AIME 2026: palindrome counting. Base uses one solver; DynaFlow connects three complementary computations to a common verifier. The dashed retry branch was not triggered. Repeated input and result packaging is omitted.

Figure 9: GPQA-Diamond: reaction-product identification. The dotted Base edge denotes call order without a data dependency; the displayed Base program contains no explicit return statement. DynaFlow shows its second invocation, integrating research evidence and verification results before answer selection.

Figure 10: HotpotQA: person identification and year retrieval. Base shows the first of three invocations. DynaFlow resolves two candidates under a top-three cap; starred calls lack valid structured outputs and are filtered out. For person candidates, resolution prompts prioritize birth-year retrieval.

Figure 11: AIME 2025: geometric optimization. Base shows its second invocation; DynaFlow shows its first. The starred optimizer was excluded before fitting because it returned no valid output. Both audits confirmed the candidate, so the dashed conditional repair call was not executed.

Multi-Agent Orchestration

Multi-agent orchestration coordinates specialized AI agents, tools, and workflows to solve complex tasks efficiently and reliably.

See all 206 papers

WorkflowOps: Learning Agent Collaboration Priors for Multi-Agent Workflow Orchestration
Oct 6, 2026Qi Cheng, Shengyu Chen, Wei Cheng +4LLM Agent Orchestration LLM Agent Workflow Optimization
Rutgers University, New Brunswick, NJ · University of Pittsburgh, Pittsburgh, PA · NEC Labs America, Princeton, NJ
Inference-Time Graph Engineering for Multi-Agent LLM Workflows
Sep 4, 2026Katherine Tieu, Dongqi Fu, Yinglong Xia +3Multi-Agent LLM Systems Multi-Agent Orchestration
Meta · University of Illinois Urbana-Champaign · Work done at Meta
RoleFlow: Self-Evolving Multi-Role Agentic Collaboration via Structural Adaptation
Oct 21, 2026Yi Ge, Pan Wang, Xuehao Cui +4Multi-Agent LLM Systems LLM Agent Workflow Optimization
University of Maryland, College Park · University of Pittsburgh · The University of Texas at Dallas

Oct 6, 2026J/K move · Enter open · S save

WorkflowOps: Learning Agent Collaboration Priors for Multi-Agent Workflow Orchestration

Qi Cheng, Shengyu Chen, Wei Cheng, Zhengzhang Chen+3

Rutgers University, New Brunswick, NJ · University of Pittsburgh, Pittsburgh, PA · NEC Labs America, Princeton, NJ

LLM Agent Orchestration LLM Agent Workflow Optimization Multi-Agent Coordination Multi-Agent Collaboration

Open paper AskPDF arXiv

Multi-agent systems are increasingly deployed for complex knowledge work, yet their orchestration layers remain largely memoryless: each new task is decomposed, assigned, and executed from scratch with no benefit from prior successful executions. We present WorkflowOps, a multi-agent workflow orchestration framework that learns agent collaboration priors from historical workflows and expands its agent pool on demand to cover new capability requirements. Our approach introduces three coupled mechanisms. First, a transition probability matrix captures pairwise agent collaboration frequencies from past workflows and applies them as soft guidance during DAG workflow construction through intra-layer ordering optimization, probability-thresholded edge suggestion, and transitive reduction for parallelism maximization. Second, a sufficiency-driven agent creation loop detects capability gaps via semantic matching scores, generates specialized agents through an LLM, and simultaneously injects them into the collaboration matrix, so that newly created agents are immediately usable with predicted collaboration priors. Third, a layered semantic matching strategy uses pre-trained sentence embeddings for fast, deterministic capability matching as a first pass, invoking LLM verification only for low-confidence cases, thereby reducing LLM routing calls by over 80% compared to pure-LLM approaches. Experiments on mixed code, math, and question-answering suites show that WorkflowOps improves end-to-end pass rates over recent workflow-construction baselines, with the largest gains on structured, decomposable tasks where past agent handoff patterns transfer.

Abstract

Multi-Agent Orchestration

Explore similar work

WorkflowOps: Learning Agent Collaboration Priors for Multi-Agent Workflow Orchestration

Inference-Time Graph Engineering for Multi-Agent LLM Workflows

RoleFlow: Self-Evolving Multi-Role Agentic Collaboration via Structural Adaptation