Not delivered · Study1views

Web Research study

With Web Research vs Baseline · 5 tasks · 2 harnesses · 2 models

Web Research does not justify its efficiency cost.

Both arms reached the same result, and the baseline got there with 36% fewer turns and 30% lower cost.

Abstract

Web Research increased costs and latency across 5 tasks over 2 harness/model cells. Cost changed +43.8%, with a median of $0.06 (mean $0.05) versus $0.04 (mean $0.05) for baseline. Token consumption changed +29.2%, reaching a median of 133k (mean 135.7k) against 103k (mean 102k). Duration increased +25.2%, with a median of 1m 00s (mean 1m 19s) compared to 48s (mean 56.6s). The treatment's worst case is worse than the baseline's: duration extended to 3m 27s against 2m 14s, and tokens reached 236.5k against 188.6k. Because harness and model are not crossed, effects cannot be attributed to either alone.

The result

BaselineWith Web Research

Best arm overall: Baseline (69.2 vs 64.0 of 100)

Same outcome on both arms — the whole gap is efficiency (cost · tokens · duration).

Best setup: claude-code / claude-haiku-4-5-20251001 · Baseline (74.2 of 100)

  1. 1claude-code / claude-haiku-4-5-20251001Baseline74.2best
  2. 2claude-code / claude-haiku-4-5-20251001With Web Research68.1
  3. 3codex / gpt-5.4-miniBaseline65.0
  4. 4codex / gpt-5.4-miniWith Web Research61.0

Overall score per arm, 0–100 points (not a pass rate): 30% goal + 30% quality + 25% efficiency (cost · tokens · duration, vs the board's best arm). Absent components renormalize. Best at the top — the board's best setup is tagged.

Head-to-head

Both arms reached the same result, and the baseline got there with 36% fewer turns and 30% lower cost.

Same result both sides — decided on what it cost

What the blind judge alone said

Baseline wins 1Ties 4Web Research wins 5Win rate 50%

In blind position-debiased comparison, Web Research won 5 of 10 pairs; 4 were ties.

Quality × efficiency clusters — normalized per task

Every graded run, standardized WITHIN its task so difficulty cancels out: → right = fewer tokens than the field on the same task, ↑ up = higher quality index (70% rubric pass rate + 15% intent + 15% outcome) than the field. The field pools every harness and model, so a setup's left–right position largely reflects its own token habits. Dots are runs; each harness logo is a harness × model × arm centroid — hover it for the model and averages. Up-right wins.

Label
Task
Arm
Harness
Model
Verdict
20 of 20 runs match
Show
BaselineWith Web Research
better · cheaperworse · pricier
← pricier than the fieldefficiency (σ, per task)cheaper than the field →

Every metric, per harness × model

Task completion

  1. claude-code / claude-haiku-4-5-20251001100% → 100%even
  2. codex / gpt-5.4-mini100% → 100%even

Goal achievement

  1. claude-code / claude-haiku-4-5-2025100160% → 60%even
  2. codex / gpt-5.4-mini40% → 40%even

Rubric quality /100

  1. claude-code / claude-haiku-4-5-2025100184 → 884.0 pts better
  2. codex / gpt-5.4-mini84 → 804.0 pts worse

Cost per run

  1. claude-code / claude-haiku-4-5-20251001$0.03 → $0.0451% worse
  2. codex / gpt-5.4-mini$0.07 → $0.0812% worse

Tokens per run

  1. codex / gpt-5.4-mini59.9k → 70.7k18% worse
  2. claude-code / claude-haiku-4-5-20251001145.8k → 225.8k55% worse

Duration per run

  1. codex / gpt-5.4-mini1m 03s → 1m 09s10% worse
  2. claude-code / claude-haiku-4-5-2025100149.4s → 1m 09s40% worse

Rubric criteria passed

  1. claude-code / claude-haiku-4-5-2025100184% → 88%4.0 pp better
  2. codex / gpt-5.4-mini84% → 80%4.0 pp worse

Baseline

No add-ons — the agent's built-in tools only.

With Web Research

No add-ons — the agent's built-in tools only.

The publisher has not described this setup — what each group's tools are, who built them, and how the tasks were chosen.

Distributions

Every completed run is one dot — the spread the averages hide. Click a dot to replay that run's journey.

Rubric quality per run

BaselineWith Web Research
code/claude-haiku-4-5-20251001
codex/gpt-5.4-mini
0100

Token composition — average per run

InputOutputCache readCache write
code/claude-haiku-4-5-20251001 · base
145k
code/claude-haiku-4-5-20251001 · treat
205.9k
codex/gpt-5.4-mini · base
59k
codex/gpt-5.4-mini · treat
65.5k

Where agents fail

Each cell is the pass fraction of one pre-registered criterion — darker red, more failures. Click a fraction to read the judge's notes on the runs behind it.

Criterionclaude-codecodexAll
T5Response identifies the 2 comparison libraries by name and cites sources that benchmark them against FastCache.0/2
T5Response cites at least 2 independent sources (e.g., published benchmark reports, GitHub issue discussions, academic papers) that compare the libraries.0/2
T1Response distinguishes between direct purchase price and cloud rental pricing, or explains why only one applies.1/2
T4Response cites the official GDPR text (e.g., EUR-Lex or the official EU GDPR website) as a source.1/2
T4Each of the 3 obligations is accompanied by a brief explanation of what it means in practice for a SaaS platform.1/2
T4Response distinguishes between what Article 25 explicitly states and any inferences about how it applies (e.g., 'Article 25 requires X; this implies Y for SaaS platforms').1/2
6 of 25 graded criteria failed at least once.

Run outcomes

How each run ended — whether the agent finished, not whether its answer passed. Whether answers passed is in the results above.

CompletedFailureTimeoutError / other
code/claude-haiku-4-5-20251001 · base
5/5 completed
code/claude-haiku-4-5-20251001 · treat
5/5 completed
codex/gpt-5.4-mini · base
5/5 completed
codex/gpt-5.4-mini · treat
5/5 completed

Efficiency frontier

Label
Task
Arm
Harness
Model
Verdict
20 of 20 runs match
Loading
BaselineWith Web Research
claude-codecodex

Statistics

MetricArmnMeanMedian (pooled)MinMaxStd dev
Rubric qualitybase10/1084906010018
treat10/1084906010018
Costbase10/10$0.05$0.03$0.02$0.12$0.03
treat10/10$0.06$0.05$0.03$0.11$0.03
Durationbase10/1056.6s50.4s26.7s2m 14s31.3s
treat10/101m 19s1m 09s32.9s3m 27s51.4s
Total tokensbase10/10102k90.9k25k188.6k59.4k
treat10/10135.7k101.2k35.4k236.5k83.8k
Output tokensbase10/103.3k2k9009.4k2.9k
treat10/104.6k3.2k1.8k11.2k3.2k
Cache-read tokensbase10/1066.8k37.7k0178.4k75.1k
treat10/1096.9k50.6k0220.7k107.8k
Cache-write tokensbase10/106k2.3k037.2k11.4k
treat10/104.8k4.2k012.4k5.2k
Turnsbase10/106.242154.9
treat10/1010.27.52208.8

n = runs carrying the fact / completed runs in the arm; every statistic runs over present facts only — a missing fact is never counted as 0. Std dev is the sample form (n−1), withheld below n = 2. Every figure here POOLS all runs. The typical-run figures in the abstract, the head-to-head decision and the per-setup panels take each task's median first and then the median across tasks, so a task that ran more often never outweighs one that ran once; the two medians can differ. A promise study's headline (“cut average tokens”) compares per-arm means.

Tasks

Runs

Every run is inspectable — open one to replay the agent's journey step by step, with the analyst's read underneath.

Every run, filterable… of 20Show runs
Task
Arm
Harness
Model
… of 20 runs match
Sort
HarnessModelTaskArmCompletedQualityPairTokensCostDuration

Loading runs…

Methodology

What each metric means
Completed
The run finished and the harness returned a response. It does NOT mean the answer was correct — a completed run can score zero on quality.
Denominator: Terminal runs, excluding those killed by our own infrastructure.
Goal achievement
The run fully achieved the task's goal: the judge panel passed every pre-registered rubric criterion (a rubric score of 100).
Denominator: Graded completed runs.
Rubric quality
The share of pre-registered acceptance criteria the judge panel passed, as a 0-100 score. Partial credit is possible.
Denominator: The criteria count frozen before any run — never the judge's returned count.
Quality index
A composite used only in the quality x efficiency plot, built from whichever grader scored this study's runs — the judge's rubric, the pre-registered checks or the benchmark's own verdict — plus whether the run's own outcome was success. The map's caption names the grader and its weights.
Denominator: Graded runs — runs the study's grader scored.
Pairwise verdict
A blind, position-debiased comparison of the two arms' final answers. It sees answer text only — cost and latency are measured separately.
Denominator: Task-cell pairs where both arms produced a response.
Infra-excluded
A run killed by our own infrastructure. It is a missing measurement, never a loss for the arm, and is excluded from every rate denominator.
Denominator: Reported as a count beside every affected panel.
A/B arms
Every task runs twice per harness × model cell — once without Web Research (baseline), once with it — on the same prompt, same model, cold start for both arms.
Pre-registered rubrics
Each task's acceptance criteria (5 per task) are written at task generation, before any run exists, so grading can never be shaped by the results.
Judge panel
A panel of independent judges (claude-opus-5, gpt-5.5, claude-fable-5), each at provider-default sampling (claude-opus-5, gpt-5.5 and claude-fable-5 do not accept a temperature setting), scores every successful response against its rubric; a criterion passes only when a strict majority of the panel passes it. Each judge sees only the task, the rubric, and the response — never which arm produced it, never token counts.
Blind pairwise
Each task's two responses are also compared blind as “Response A” and “Response B” by every panel judge, each judging twice with the order swapped (disagreement = tie); the pair's verdict is the panel's strict majority — no majority counts as a tie.
How the head-to-head is decided
The conclusion is not the blind judge's alone — that judge sees only the two response texts, never the rubric score or what each run cost. Evidence is ranked in order: whether the task's goal was achieved, then the rubric score, then the measured cost of getting there (tokens, turns, duration, spend), and only then the blind judge, for pairs nothing else separates. Cost never outranks correctness, and when neither arm achieved the goal the efficiency gap between them decides nothing.
One run per task
Every task ran once per cell and arm, in one phrasing. Run-to-run variation is therefore not measured: a single task's difference can be one lucky or unlucky attempt. Every row of the reference table treats the TASKS as the unit: its intervals resample the tasks and its p-values come from a paired test over them, so they describe how much the result depends on which tasks were drawn — not how it would change if the same tasks were run again. With few tasks no difference can reach significance (five tasks cannot go below p = 1/16).
Sample size
2 cells × 5 tasks × 2 arms = 20 runs.