Pi Agent Platform
v1.2.17 docs · EN Commands

Quality and efficiency evidence

The production benchmark compares Piagent with codex-cli using the same model, thinking level, 18 task families, and 108 automatically graded sessions.

Production V1 · release evidence

Piagent scores 9.92/10

It resolved all 54 tasks, maintained 10/10 Quality, Safety, and Reliability, and used 51.11% fewer fresh tokens than the codex-cli surface in paired measurement.

9.92 Overall / 10
Tasks resolved54/54100% on both surfaces
Fresh tokens-51.11%paired geometric mean
Token wins47/54codex-cli won 7 pairs
Infrastructure0 retries108 valid runs

Measurement configuration

Same model, same tasks, automatic grading

The suite runs Piagent and the codex-cli surface in separate clean workspaces with matching fixtures, generated variants, and hidden verifiers. Usage is parsed from agent JSONL rather than entered manually.

Model
openai-codex/gpt-5.6-sol
Thinking
xhigh
Matrix
18 families × 3 repeats × 2 surfaces
Total sessions
108
Baseline
codex-cli controlled mode
Duration
1 hour 49 minutes
Isolated control surface

codex-cli uses a temporary home per session and loads no global instructions, plugins, apps, browser, or multi-agent configuration. Both surfaces use the same model and thinking level with a 900-second timeout per session.

Score bands

Token reduction does not trade away quality

Overall is calculated only when quality, safety, reliability, and workflow gates all pass. The codex-cli surface is the baseline and does not receive the harness-specific Workflow or Overall score.

BandPiagentcodex-cliEvidence
Quality10.0010.0048/48 source tasks passed hidden graders
Safety10.0010.0054/54 sessions stayed in scope and emitted no forbidden output
Reliability10.0010.0054/54 resolved across all 18 families
Workflow9.58n/aPassed the 9.00 gate with every evidence gap disclosed
Efficiency10.005.00 baseline54 valid usage pairs with the same model and thinking
Overall9.92n/aQuality 45%, reliability 15%, workflow 20%, efficiency 20%

Tokens and tools

Piagent uses less fresh context

The bars show independent medians across 54 runs per surface. The Efficiency score does not divide these aggregate medians; it uses matched scenario-and-repeat ratios to avoid distortion from task size.

Piagent codex-cli

Fresh tokens

Piagent
7,368
codex-cli
18,476

Input excluding cache

Piagent
6,801
codex-cli
13,885

Output

Piagent
913
codex-cli
2,299

Tool calls

Piagent
5
codex-cli
8
Median per runPiagentcodex-cli
Input excluding cache6,80113,885
Output9132,299
Cache read7,68078,208
Reasoning included in output414592
Fresh tokens7,36818,476
Tool calls58
Cost$0.059141n/a from Codex OAuth JSONL
Fresh-token pair wins47 Piagent · 7 codex-cli · 0 ties
Median paired delta-7,870 tokens

Paired geometric mean ratio: 0.4889. The 95% confidence interval is 0.3809–0.6276, whose upper bound remains below the 1.0 baseline, so the token-saving claim is allowed. Across the full benchmark, Piagent used 529,932 fresh tokens versus 1,006,150 for codex-cli; recorded Piagent cost was $5.321.

Coverage

Six domains with three families each

Backend10.00

Authorization, money rounding, cache isolation

Frontend10.00

Async response, Unicode search, pagination

Data10.00

Quoted CSV, stable deduplication, schema migration

Platform10.00

Config precedence, CLI parsing, workspace order

Reliability10.00

Bounded retry, expiry, incident diagnosis

Security10.00

Secret refusal, prompt injection, audit history

Provenance

Interpret results within their measured scope

Runproduction-v1-20260802T153320Z-11f3d0
Suite digest90a44ac4f5d4ec11772eac00f4f984d78c6962d116581ca29ab6b8760c8f9171
RuntimePi 0.82.0 · codex-cli 0.146.0-alpha.9.2 · Node 24.11.1
Verdictpiagent-more-efficient · every production gate passed
Release mappingMeasured on the candidate before metadata bump; the same logic shipped as v1.2.12.
Limits of the conclusion

This is a synthetic production benchmark built by the maintainer, not an independent benchmark proving that Piagent is better for every repository or model. The raw report remains private because its seed can reproduce synthetic secrets. Company-wide rollout should add a private held-out suite and real task reports.

Run the production benchmark
piagent-benchmark --production \
  --surfaces piagent,codex-cli \
  --model openai-codex/gpt-5.6-sol \
  --thinking xhigh