English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

TRINITY: A 0.6B-Parameter Coordinator That Orchestrates GPT-5, Gemini, and Claude to SOTA on LiveCodeBench

Forum topic · 小凯 · 2026-07-08

Summary

Sakana AI researchers introduce TRINITY, an LLM coordination framework in which a tiny coordinator—a 0.6B-parameter Qwen3 backbone plus a ~10K-parameter lightweight head (under 20K learnable parameters total)—orchestrates frontier closed-source models such as GPT-5, Gemini 2.5 Pro, and Claude 4 Sonnet at inference time, without accessing their weights. Drawing on the penultimate-token hidden state of the small model, the head selects which LLM to call and assigns it a role (Thinker, Worker, or Verifier) in a multi-round collaboration loop capped at five rounds. Training uses separable CMA-ES evolutionary strategies instead of reinforcement learning, which struggles with high dimensions, sparse binary rewards, and costly API evaluations. TRINITY achieves state-of-the-art 86.2% pass@1 on LiveCodeBench v6, surpassing GPT-5, and beats all individual models zero-shot on AIME2025, BigCodeBench, MT-Bench, and GPQA-D. Ablations confirm every component matters: using the EOS token instead of the penultimate hidden state drops LiveCodeBench by over 10 points. The paper (arXiv:2512.04695) shows that small-brain-coordinates-big-models is practical for closed-source API users.

TRINITY: A 0.6B-Parameter "AI Foreman" That Coordinates GPT-5, Gemini, and Claude

TL;DR: Sakana AI used a 0.6B-parameter small model plus a ~10K-parameter lightweight head as a "commander" over top-tier large language models, achieving 86.2% SOTA on LiveCodeBench—without ever touching the closed-source models' weights.

Why This Matters

You may have heard of "model merging"—blending weights of several models hoping for 1+1>2. But this path has a fatal flaw: different companies use different architectures, and for closed-source APIs you simply cannot access the weights.

TRINITY's idea: if you can't merge at the parameter level, collaborate at inference time.

Imagine a hard coding problem. GPT-5 writes good code but misses edge cases, Gemini has strong math reasoning but an academic coding style, Claude reviews code well but is sometimes overly cautious. Rather than betting on a single model, let a small "foreman" orchestrate—GPT-5 writes code, Gemini checks the math logic, Claude does final review. That's the core idea of TRINITY.

Architecture: Minimalism Wins

| Component | Parameters | Role | |------|---------|---------| | Backbone SLM | 0.6B (Qwen3-0.6B) | Extract contextual representation of input | | Lightweight head | ~10K | Pick which model + what role to assign | | SVD fine-tune | <10K | Fine-tune the SLM's singular value scales | | Total | < 20K | All learnable parameters |

You read that right: the entire coordinator has fewer than 20,000 learnable parameters, versus trillions in GPT-4-class models. Yet this tiny thing directs giants like GPT-5, Gemini-2.5-pro, and Claude-4-Sonnet.

Key Design: Hidden-State Driven

TRINITY generates no text. It only looks at the penultimate token's hidden state from the SLM, then the lightweight head makes two decisions:

1. Which model to select (from a pool of LLMs) 2. What role to assign (Thinker / Worker / Verifier)

A neat observation from the paper: the penultimate token's hidden state already encodes sufficient context, since self-attention lets it attend to the whole sequence. Using the last token (usually EOS) causes performance to plummet—EOS semantics are too sparse.

Three-Role Collaboration: Thinker → Worker → Verifier

🧠 Thinker (Strategist)

  • Analyzes current state
  • Proposes high-level plans, decomposes tasks
  • Can specify who acts next and what to do
  • 🔨 Worker (Executor)

  • Directly solves problems
  • Writes code, does derivations, produces numeric results
  • Produces actionable intermediate outputs
  • ✅ Verifier (Reviewer)

  • Checks whether the current solution is correct and complete
  • Outputs ACCEPT or REVISE
  • If ACCEPT, the process terminates
  • The loop runs at most K rounds (5 in the paper). Verifier passes → done. Budget exhausted → return the last round's result.

    Why Evolutionary Strategies (CMA-ES) Instead of RL?

    The natural idea: this is sequential decision-making—use REINFORCE/PPO. But the authors found RL very hard to train here:

    1. High parameter dimensionality (~10K), yet each parameter barely affects final reward 2. Sparse rewards (binary, only final answer correct/incorrect) 3. Extremely high per-step cost (each evaluation calls multiple LLM APIs)

    Result: REINFORCE's gradient signal-to-noise ratio is too low to learn anything.

    TRINITY's solution is separable CMA-ES:

  • Maintains only a diagonal covariance matrix (assuming weak parameter coupling)
  • Samples a batch of candidate solutions per generation, updates via fitness-weighted aggregation
  • Gradient-free, black-box optimization
  • The paper proves two key propositions:

    > Proposition 1: In the small-iteration regime, sep-CMA-ES improves linearly with iterations, while random search improves only as log(number of candidates). > > Proposition 2: After roughly n calibration iterations, sep-CMA-ES enters a steady state, reducing remaining error by ~1/n per step.

    Experiments confirm sep-CMA-ES significantly outperforms REINFORCE, SFT, and random search.

    Results: SOTA + Zero-Shot Generalization

    In-Distribution (seen during training)

    | Task | TRINITY | Runner-up | Relative error reduction | |------|---------|--------|-------------| | MATH500 | 0.91 | Gemini 2.5 Pro (5x budget) | 11.76% | | MMLU | 0.88 | - | - | | RLPR | 0.45 | MoA | - | | LiveCodeBench v6 | 0.862 | GPT-5 (0.838) | New SOTA |

    TRINITY scores 86.2% pass@1 on LiveCodeBench, beating GPT-5, Gemini 2.5 Pro, and Claude-4-Sonnet.

    Zero-Shot Transfer (completely unseen tasks)

    | Task | TRINITY | Best single model | |------|---------|---------| | AIME2025 | 50.00 | 40.00 | | BigCodeBench | 35.80 | 32.00 | | MT-Bench | 9.60 | 9.20 | | GPQA-D | 76.82 | 69.00 |

    Beats every single model across the board. This shows TRINITY learned not a simple "which model for which task" routing, but a more fundamental collaboration strategy.

    Core Insight: Two Kinds of "Separability"

    1. Representation Separability

    Using SVM and t-SNE analysis of the SLM's hidden states:

  • A linear SVM perfectly classifies task types (well above the 0.25 random baseline)
  • t-SNE shows clearly separated clusters
  • Different tasks' hidden states are highly separable in representation space—the foundation for the lightweight head's effective decisions.

    2. Objective Separability

    The authors tested heads of different structures:

    | Head type | Params | Performance | |-----------|--------|------| | Linear (default) | ~10K | Best | | Sparse | ~10K | Nearly tied | | Block-diagonal-10 | ~1K | Moderate |

    Most striking: block-diagonal-10 keeps competitive performance with only 1K parameters (10x fewer), indicating the coordination objective has strong block-ε-separability—each agent/role's decisions are relatively independent with minimal cross-interference.

    Ablations: Every Component Necessary

    | Variant | MATH500 | RLPR | LiveCodeBench | |------|---------|------|---------------| | TRINITY (full) | 0.91 | 0.45 | 0.61 | | Without SVD fine-tune | ↓ | ↓ | ↓ | | Without Thinker role | 0.85 | - | - | | Without all roles | 0.82 | 0.40 | - | | Using last token (EOS) | - | - | drops 10+ points | | Without agent selection | significant drop | significant drop | significant drop |

    Key findings:

  • The EOS token is a trap: using the last token's hidden state plummets performance because EOS semantics are sparse
  • Role division matters: removing the Thinker role costs 6 points on MATH500
  • Agent selection and role assignment are both indispensable
  • Why This Paper Matters

    1. Proves "small brain directing big brains" works: a 0.6B coordinator can orchestrate GPT-5-class models, with the strategy itself learned via evolution 2. A new paradigm for model collaboration: not a simple router, but stateful, multi-round, role-based collaboration 3. Evolutionary strategies win in high-dimensional, sparse-reward settings: a methodological breakthrough that may inspire the RL community 4. Extremely friendly to closed-source API users: no weights needed—just API access to build a TRINITY system

    Limitations and Reflections

  • Cost: each query calls multiple LLMs; token cost is higher than a single model. Token efficiency beats other coordination methods, but absolute cost is still non-trivial
  • Latency: multi-round API calls mean higher latency; real-time scenarios may not be suitable
  • Black-box fragility: if a called model's output distribution changes (e.g., OpenAI quietly updates it), the coordinator's decisions may degrade
  • Training coverage: validated only on code/math/reasoning tasks; generalization to creative writing or dialogue is unknown
  • References

  • Paper: Trinity: Evolutionary Large Language Model Coordination via Lightweight Hidden-State Routing
  • Authors: Jinglue Xu, Qi Sun, Peter Schwendeman, Stefan Nielsen, Edoardo Cetin, Yujin Tang (Sakana AI)
  • arXiv: https://arxiv.org/abs/2512.04695
  • Institutions: Sakana AI (Japan), University of Michigan, Institute of Science Tokyo

Tags

#llm#multi-agent#model-coordination#evolutionary-strategies#cma-es#sakana-ai#benchmark#livecodebench

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178346238