Back to Ruflo

Performance SOTA Report — 2026-07-20

docs/dream-cycles/2026-07-20-performance-sota.md

3.34.05.2 KB
Original Source

Performance SOTA Report — 2026-07-20

TL;DR: World-model agent planners achieve 14× training speedup + 3–6× inference speedup in 2026 (Grade A), while the Ruflo Flash Attention claim remains unverified and framework throughput is unmeasured vs competitors delivering 84% more rps.


What's New in 2026

FindingSourceConfidence
DSWorld world-model: 14× RL agent training speedup, 3–6× search inference speedup by simulating tool outcomesarXiv:2607.15901 (ICRA-style 2026 preprint)A
AutoAgents-Rust: 4.97 rps vs LangGraph Python 2.70 rps — 84% throughput gain from static-typed Rust orchestrationdev.to/saivishwak (2026 benchmark, cross-checked with n1n.ai report)B
Framework choice moves agent benchmark performance by up to 30 percentage points on identical modelstensoria.fr/en/blog (2026, qualitative)C
SEED self-evolving distillation: dense token-level supervision from hindsight skills improves performance + sample efficiency on agentic tasksarXiv:2607.14777 (2026 preprint)A
LangGraph P95 latency: 16,891 ms; orchestration adds ~50–100 ms over raw LLM calls (5.7–7 s)mcp-server-langgraph.mintlify.app benchmarks + aerospike.comB
VideoTreeSearch (VTS): self-correcting hierarchical search +12.5 mIoU via explicit backtracking over prior agentic baselinesarXiv:2607.16189 (2026)A
CrewAI carries 3× token overhead of LangGraph on simple workflowstopuzas.medium.com / dev.to (2026 showdown)C

Ruflo Current Capability

AreaRuflo State
Flash Attention speedup2.49×–7.47× claimed — explicitly unverified in CLAUDE.md; no benchmark exists
MCP response target<100 ms — target, not measured
CLI startup target<500 ms — target, not measured
Framework throughputNot benchmarked vs LangGraph / AutoGen / CrewAI
Agent task planningReactive execution only — no world-model pre-simulation
Token efficiency3-tier routing (ADR-026/ADR-143) but savings vs competitors unmeasured
HNSW searchMeasured ~1.9× at N=20k, ~3.2×–4.7× at N=5k — verified

Competitor Comparison

FrameworkThroughput (rps)P95 LatencyToken OverheadWorld-Model Planning
RufloNot benchmarkedNot benchmarked3-tier routing (B)None
LangGraph2.70 rps (B)16,891 ms (B)BaselineNone
AutoAgents (Rust)4.97 rps (B)Not reportedLower than LangGraph (C)None
AutoGenSlowest of four (C)Not reportedHigh consensus overhead (C)None
CrewAINot reportedNot reported3× vs LangGraph (C)None
OpenAI SwarmLowest latency (C)Not reportedLowest overhead (C)None

All competitor figures are 2026 third-party benchmarks. "C" = single source, unconfirmed.


Benchmarks

BenchmarkValueGradeSource
DSWorld 14× RL training speedupMeasured on 8,000-scale transition trajectory dataset vs LLM baseline (+35.6% transition prediction)AarXiv:2607.15901
DSWorld 3–6× search inference speedupSame paper, search-based inference pathAarXiv:2607.15901
AutoAgents-Rust 4.97 vs LangGraph 2.70 rpsCross-checked dev.to + n1n.ai 2026 reportsBdev.to/saivishwak
LangGraph orchestration overhead~50–100 ms for complex workflowsBmcp-server-langgraph.mintlify.app
Ruflo Flash AttentionNo 2026 benchmark data available — claim must be labeled unverified until measuredCLAUDE.md audit

SOTA Proof & Witness

Session commit: 12ede21767a6dd669df1b79392a5d27d9154f237
Report SHA-256: 0faf7a674d8a74ad9e7c309b8b9e0e8db9ca683cb5bfc752f2d8fd0b3899713f
Witness stamp: 761ca053e5b55bdbf5205e7da61c61bdb8758d8659596f461b341346746d5d1c

Verifier: sha256sum /path/to/dream-gist-2026-07-20.md → concat with session commit → sha256sum → must equal Witness stamp.


  1. Implement a lightweight world-model planner for Ruflo agents (ADR-322): before executing expensive multi-step tool chains, agents simulate outcomes using a cost-aware routing + transition predictor (DSWorld pattern). Target: 3–6× reduction in wasted real-tool calls during planning. Scope: new @claude-flow/planning module; hooks pre-task triggers planner when task has ≥3 estimated tool calls.

  2. Run a Ruflo throughput benchmark vs LangGraph (one sprint): instrument claude-flow start with autocannon or wrk targeting the MCP stdio/HTTP layer; publish results. Current unverified Flash Attention and <100 ms MCP claims become Grade A or are retracted. Without this, marketing claims are Grade C.

  3. Compress SendMessage payload with token budget enforcement (security × performance overlap): each SendMessage call should enforce a configurable token budget (default 512 tokens) and optionally apply a lightweight compression pass (TF-IDF keyword extraction). Addresses OWASP ASI07 (inter-agent channel) and reduces orchestration overhead on busy swarms.