Measured · Study1views

acceptance row1 1790522824583

GSM8K · 3 tasks · 1 harness · 1 model

GSM8K augmentation delivered the strongest efficiency gains with openai-compatible / gpt-oss-120b.

Ranked shootout, no baseline arm: cells ranked by the composite score — the benchmark's own graders, then efficiency — openai-compatible / gpt-oss-120b leads.

Abstract

This single-arm shootout evaluated GSM8K augmentation across 3 tasks and 2 harness/model cells, with openai-compatible / gpt-oss-120b as the leader. Efficiency was the deciding factor: the treatment achieved a median cost of $0.0001 (mean $0.0001), median tokens of 277 (mean 330), and median duration of 719ms (mean 994ms).

The result

Color by

openai-compatible

Best setup: openai-compatible / gpt-oss-120b · effort high (99.7 of 100)

  1. 1openai-compatible / gpt-oss-120b · effort high99.7best
  2. 2openai-compatible / gpt-oss-120b · effort low72.6

Overall score per cell, 0–100 points (not a pass rate): 75% benchmark pass rate (the benchmark's own graders carry the outcome share — no judge grades or evals on this board) + 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.

Leaderboard

Cells ranked by the composite score — outcome (the lenses that graded) × efficiency — absolute values, no baseline arm. Tokens, cost and duration are per-cell medians, the efficiency the score folds.

#HarnessModelScorePass rateQualityEvalsGoalTokensCostDurationRuns
1openai-compatiblegpt-oss-120b · effort high99.7100%−−100%254$0.0001743ms3/3
2openai-compatiblegpt-oss-120b · effort low72.667%−−100%299$0.0001834ms3/3

Pass rate

83.3% (95% CI 50–100)over 6 graded runs
CellPass rateTasksRunsRan asMatcher
openai-compatible/accounts/fireworks/models/gpt-oss-120b · effort high100%33as publishednumeric
openai-compatible/accounts/fireworks/models/gpt-oss-120b · effort low66.7% (95% CI 0–100)33as publishednumeric

Each run's answer was extracted from its response and matched against the benchmark's own key — no model in the loop. Two-stage: the mean over repeats per task, then the mean over tasks, with the interval taken over tasks.

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 (85% the benchmark's own pass/fail + 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.

Task
6 of 6 runs match
Show
Color by

openai-compatible

better · cheaperworse · pricier
← pricier than the fieldefficiency (σ, per task)cheaper than the field →

Every metric, per harness × model

Benchmark pass rate

  1. openai-compatible/accounts/fireworks/models/gpt-oss-120b · effort high100%
  2. openai-compatible/accounts/fireworks/models/gpt-oss-120b · effort low67%

Task completion

  1. openai-compatible/accounts/fireworks/models/gpt-oss-120b · effort high100%
  2. openai-compatible/accounts/fireworks/models/gpt-oss-120b · effort low100%

Cost per run

  1. openai-compatible/accounts/fireworks/models/gpt-oss-120b · effort high$0.0001
  2. openai-compatible/accounts/fireworks/models/gpt-oss-120b · effort low$0.0001

Tokens per run

  1. openai-compatible/accounts/fireworks/models/gpt-oss-120b · effort high254
  2. openai-compatible/accounts/fireworks/models/gpt-oss-120b · effort low299

Duration per run

  1. openai-compatible/accounts/fireworks/models/gpt-oss-120b · effort high743ms
  2. openai-compatible/accounts/fireworks/models/gpt-oss-120b · effort low834ms

By harness, by model

Benchmark pass rate

By harness

  1. openai-compatible83%

By model

  1. gpt-oss-120b83%

Task completion

By harness

  1. openai-compatible100%

By model

  1. gpt-oss-120b100%

Cost per run

By harness

  1. openai-compatible$0.0001

By model

  1. gpt-oss-120b$0.0001

Tokens per run

By harness

  1. openai-compatible277

By model

  1. gpt-oss-120b277

Duration per run

By harness

  1. openai-compatible719ms

By model

  1. gpt-oss-120b719ms

Distributions

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

Cost per run

Your setup
openai-compatible/gpt-oss-120b · effort high
openai-compatible/gpt-oss-120b · effort low
0$0.0003

Token composition — average per run

InputOutputCache readCache write
openai-compatible/gpt-oss-120b · effort high · treat
329
openai-compatible/gpt-oss-120b · effort low · treat
331

Statistics

MetricArmnMeanMedian (pooled)MinMaxStd dev
Costtreat6/6$0.0001$0.0001$0.0001$0.0003$0.0001
Durationtreat6/6994ms789ms563ms1.7s493ms
Total tokenstreat6/6330277207525140
Output tokenstreat6/619412491386135
Turnstreat6/611110

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.

What each setup did

Derived from each run's recorded tool calls — not from a model's description of the run — and aggregated per setup, so a behavior seen across several runs is stated once with its rate. Open a finding to see the runs behind it, each linked to its journey at the step where it happened.

Task
Pick a finding or a setup to list the runs behind it.
openai-compatible · fireworks · gpt-oss-120b · Your setup1 finding
  • 6 of 6 runs · 6 completed

Task text withheld

Task text withheld — this benchmark is guarded and its tasks are not republished here.

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 6Show runs
Task
… of 6 runs match
Sort
HarnessModelTaskCompletedPassTokensCostDuration

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.
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.
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.
Single arm
Every task runs once per harness × model cell — a shootout with no baseline arm. The readout is absolute (quality, success, tokens, cost) and the cells are ranked into a leaderboard.
One run per task
Every task ran once per cell, 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 × 3 tasks × 1 arm = 6 runs.