Skip to content
Back to research
Benchmark report

The model is only one part of the agent.

Dialogo improved every model evaluated to date by 8-12 percentage points on autonomous tool-use tasks, without fine-tuning or changing model weights.

74.4%Dialogo + Gemini 3 Flash
+10.4 ppvs. published baseline
500public benchmark tasks
0model weights changed
Model-agnosticTool usePublic-500Methodology disclosed

The result

A stronger harness changed the outcome. The model stayed the same.

On MCP-Atlas Public-500, Dialogo with Gemini 3 Flash reached approximately 74.4%, compared with the published Gemini 3 Flash baseline of 64.0%.

That is a 10.4 percentage-point uplift, or 16.25% relative improvement.

MCP-Atlas Public-500

Pass rate with the same executor

+10.4 pp
0%20%40%60%80%100%Published baseline64.0%With Dialogo74.4%+10.4 points

The direct pass-rate difference is 10.4 percentage points. Expressed relatively, the increase is 16.25%.

The wider field

A smaller model can compete in a bigger-model performance band.

With Dialogo, Gemini 3 Flash moves from its 64.0% published baseline to 74.4%. In this evaluation, that puts the efficient model in the same performance band as several substantially larger frontier models.

Public-500 comparison

Dialogo in the full published field

Dialogo runScale leaderboard
0%20%40%60%80%100%Muse Spark82.4Claude Opus 4.7 (max)80.8Gemini 3.1 Pro Preview (high)80.6GPT-5.5 (xhigh)79.4Claude Opus 4.6 (max)79.0GLM-5.178.6GPT-5.4 (xhigh)75.4Dialogo + Gemini 3 Flash74.4Claude Opus 4.5 (high)73.4Gemini 3 Pro Preview73.0Claude Sonnet 4.672.8GPT-5.2 (xhigh)71.8Kimi K2.564.0Gemini 3 Flash Preview64.0Claude Sonnet 4.5 (thinking)62.0GLM-4.761.2GPT-5.4 Mini (xhigh)61.0Gemini 3.1 Flash Lite (high)60.4GPT-5.1 (high)53.6o3-pro48.8Claude Haiku 4.541.2
Official model values are Scale's Public-500 scores. The highlighted 74.4% is Dialogo's independently measured result and is not an official Scale leaderboard submission. Different harness and configuration choices can affect results.

What Dialogo is

A model-independent execution layer

A frontier, efficient, specialized, or self-hosted open-weight model can remain the executor. The model can be replaced without rebuilding the complete workflow.

An agent harness around the model

Dialogo manages planning, tool use, context, recovery, state, verification, permissions, and human approvals without changing model weights.

Where the uplift comes from

Intelligence is a property of the whole execution system.

01

Plan and track subgoals

Break a goal into accountable steps, preserve execution state, and detect incomplete requirements.

02

Use tools with control

Discover tools, generate schema-aware arguments, manage outputs, and recover from transient failures.

03

Verify before delivery

Check intermediate results, validate the final answer, and pause consequential actions for approval.

04

Optimize successful work

Route by complexity, cap unnecessary calls, and measure cost against completed tasks instead of tokens alone.

Foundation modelReplaceable executorPlanSelect toolsVerifyOperational environmentTools + dataPermissions + approvalsExecution historyDIALOGO HARNESS

The model produces decisions. The harness controls how those decisions become reliable actions.

The efficiency question

Measure cost per successful task, not cost per token alone.

A smaller model is only economical if it completes the work. A stronger harness can route simple steps efficiently, reserve expensive reasoning for difficult decisions, limit oversized outputs, and recover without restarting the complete task.

Total execution costModels + tools + runtime/Passed tasksVerified outcomes only=Cost per successThe operating metric
Total model cost
Pass rate
Tool-call count
p95 task latency

Research context

Agent architecture transfers across models.

Sakana AI's Fugu explores learned model orchestration through selection, delegation, verification, and synthesis. Darwin Gödel Machine and ShinkaEvolve provide further evidence that improvements to an agent scaffold can generalize across underlying models.

System-level intelligence depends not only on the model, but on how models, tools, context, verification, and execution are orchestrated.

Methodology and evidence

What the current evaluation shows.

MCP-Atlas evaluates realistic, multi-step tool use. The public split contains 500 tasks across real MCP servers, and tasks pass when claim coverage reaches the benchmark threshold.

View Scale Labs benchmark

Measured results

  • Every model evaluated by Dialogo to date improved.
  • Observed gains range from 8 to 12 percentage points.
  • Gemini 3 Flash showed a 10.4-point uplift against the cited public baseline.
  • No underlying model weights were modified.

Operational implication

The measured Gemini 3 Flash configuration reached a performance band occupied by larger frontier models. This supports a practical strategy: improve the orchestration system first, then select the smallest executor that meets the workflow's quality, latency, privacy, and cost requirements.

Comparison note: Scale updated and re-scored its leaderboard methodology in April 2026. Official comparison values shown above use the published Public-500 column. Dialogo's approximately 74.4% result is independently measured. Full task-level traces, model configurations, cost, and latency should accompany a formal reproduction.

A model call is not an autonomous system.

The model matters. So do the plan, tools, context, recovery, permissions, and verification that turn a response into completed work.