Navigator Bench · Index v2.0 · October 2026 (demo)

Demo data · not real results

Top-rank quality at 1/28 the cost per successful task

6 models share rank 1 (none is significantly better than another on the same tasks); Model M · low effort (Index 94.2) costs $0.0042 per successful task, Model D · high effort (Index 95.5) $0.12.

Navigator Index (0–100) — higher is better
Cost per successful task (log scale) — lower is better
Navigator Bench · Index v2.0 · data through 14 Oct 2026Up is better quality, left is cheaper, colour is provider.Demo data

Data through 14 Oct 20261,800 runsIndex v2.0ChangelogMethodChart card 16:91:1

02Leaderboard

Pick a category. Watch the order change.

Top-5 card

Ordered by rank. Ranks are ranges: 2–4 means we can't statistically separate it from the models in that range (paired test on the same tasks; the chance of any false win on the board is at most 5%). The bar under each rank shows where its range sits.

Model
Categories
Actions
1–6
Model DProvider 2 · high effort
95.593.8–97.0
$0.12
26 s
=+1.0
1–6
Model DProvider 2 · low effort
94.492.7–96.1
$0.045
11 s
=+0.8
1–6
Model MNEWProvider 5 · low effort
94.292.2–96.0
$0.0042
6.8 s
NEW
1–6
Model IProvider 4 · low effort
93.892.0–95.4
$0.030
9.1 s
=+0.8
1–6
Model AProvider 1 · low effort
93.091.3–94.6
$0.051
9.3 s
=±0.0
1–6
Model FProvider 3 · low effort
93.091.5–94.5
$0.026
7.9 s
=−0.7
7–9
Model KProvider 5 · low effort
83.679.6–87.4
$0.0044
7.0 s
▼1+3.8
7–9
Model BProvider 1 · low effort
78.975.0–82.6
$0.015
5.1 s
▼1−0.5
7–9
Model GProvider 3 · low effort
76.971.4–82.4
$0.0069
3.1 s
▼1+3.1
10–14
Model LProvider 5 · low effort
54.849.4–60.4
$0.0034
3.6 s
▼1+5.8
10–14
Model CProvider 1 · low effort
53.348.2–58.7
$0.0098
2.6 s
▼1+2.6
10–14
Model EProvider 2 · low effort
49.743.9–55.1
$0.010
2.3 s
▼1+2.5
10–14
Model JProvider 4
48.543.3–54.7
$0.016
2.3 s
▼1+2.0
10–14
Model HProvider 3
44.838.6–50.9
$0.0058
1.5 s
▼1+2.0
Reference · unranked, never counted in ties or dominance
REF
Human baselineHuman reference
88.084.5–91.2
$0.97
4.0 min
unranked

Navigator Bench · Index v2.0 · data through 14 Oct 2026Whiskers: 95% task-clustered intervals · How ties work · Cost per success

03Kill zone

Better and cheaper, or it doesn't count

Model M · low effort dominates 7 of 13 models on Navigator Bench — higher Index, lower cost per successful task.

Showing the largest kill zone this cycle. Pick any model here, from a point's card, or with the ⊘ button in the leaderboard; the scatter and the table dim with it.

Download card

A kill needs both: significantly better on the Index (a paired test on the same tasks, with the chance of any false win on the board held to 5%) and cheaper per successful task. A gap inside the noise never counts. How dominance is decided

  1. 1–6Model D · high95.5 · $0.12
  2. 1–6Model D · low94.4 · $0.045
  3. 1–6Model M · low94.2 · $0.0042
  4. 1–6Model I · low93.8 · $0.030
  5. 1–6Model A · low93.0 · $0.051
  6. 1–6Model F · low93.0 · $0.026
  7. 7–9Model K · low83.6 · $0.0044dominated
  8. 7–9Model B · low78.9 · $0.015dominated
  9. 7–9Model G · low76.9 · $0.0069dominated
  10. 10–14Model L · low54.8 · $0.0034
  11. 10–14Model C · low53.3 · $0.0098dominated
  12. 10–14Model E · low49.7 · $0.010dominated
  13. 10–14Model J48.5 · $0.016dominated
  14. 10–14Model H44.8 · $0.0058dominated

04Commoditization curve

What Index 70 costs, cycle by cycle

Curve card

Cheapest at Index ≥ 70 · October 2026 (demo)

$0.0042

per successful task · Model M · low effort

▼ 27% since August 2026 (demo)

Cheapest cost per successful task at Index ≥ 70 (log scale) — lower is better
Published cycle (data cut-off)
Navigator Bench · Index v2.0 · data through 14 Oct 2026Each point: the cheapest model with Index ≥ 70 that cycleDemo data

October 2026 (demo): Model M · low effort was the cheapest model at Index 70 or better, at $0.0042 per successful task. Data through 14 Oct 2026.

05Compare

Two or three models, every category

CodingData analysisExtractionReasoningResearch & agenticSummarizationSupportWriting255075100
  • Model G · low effort
  • Model D · high effort
  • Shaded rings: 95% intervals
Model G · low effort
76.9Index · 71.4–82.4 · rank 7–9
Cost / success
$0.0069
Median latency
3.1 s
Pass rate
72% (58%–83%)
Passes every time
60%
Model D · high effort
95.5Index · 93.8–97.0 · rank 1–6
Cost / success
$0.12
Median latency
26 s
Pass rate
100% (91%–100%)
Passes every time
100%
Category · weightModel GModel D
Coding 15%72.549.2–90.2 · 5 tasks97.593.8–100.0 · 5 tasks
Data analysis 15%85.878.4–93.3 · 5 tasks96.793.9–99.5 · 5 tasks
Extraction 10%88.373.4–96.7 · 5 tasks93.384.9–99.9 · 5 tasks
Reasoning 15%64.246.5–82.8 · 5 tasks95.893.0–98.6 · 5 tasks
Research & agentic 15%For now these are single-call multi-step tasks: one model call returns plan → steps → deliverable, checked step by step. Tool-using, browsing and long agentic tasks are deferred.77.568.2–84.0 · 5 tasks97.595.6–99.4 · 5 tasks
Summarization 10%65.841.6–89.1 · 5 tasks90.880.6–100.0 · 5 tasks
Support 10%76.763.6–89.7 · 5 tasks94.288.6–99.8 · 5 tasks
Writing 10%88.381.8–93.9 · 5 tasks95.090.3–99.7 · 5 tasks
Run receipts120 demo runs · no receipts120 demo runs · no receipts

Per-task receipts (prompt, output, checks and cost for every run) open from each model's receipts link.

06Fit-finder

The cheapest model that clears your bar

Successful tasks a month

Cheapest at Navigator Index ≥ 70

Model M · low effort

$41.91a month · $0.0042 per successful task

Cheapest model at Navigator Index 70.0 or better: Model M · low effort, $0.0042 per successful task (about $41.91 for 10,000 successful tasks a month).

  1. Model M · lowNavigator Index score 94.2monthly cost $41.91/mo, clears the bar
  2. Model K · lowNavigator Index score 83.6monthly cost $44.18/mo, clears the bar
  3. Model G · lowNavigator Index score 76.9monthly cost $68.89/mo, clears the bar
  4. Model B · lowNavigator Index score 78.9monthly cost $155/mo, clears the bar
  5. Model F · lowNavigator Index score 93.0monthly cost $258/mo, clears the bar
  6. Model I · lowNavigator Index score 93.8monthly cost $296/mo, clears the bar
  7. Model D · lowNavigator Index score 94.4monthly cost $447/mo, clears the bar
  8. Model A · lowNavigator Index score 93.0monthly cost $512/mo, clears the bar
  9. Model D · highNavigator Index score 95.5monthly cost $1,178/mo, clears the bar
  10. Model L · lowNavigator Index score 54.8monthly cost $34.03/mo, below the bar
  11. Model HNavigator Index score 44.8monthly cost $57.86/mo, below the bar
  12. Model C · lowNavigator Index score 53.3monthly cost $98.05/mo, below the bar
  13. Model E · lowNavigator Index score 49.7monthly cost $104/mo, below the bar
  14. Model JNavigator Index score 48.5monthly cost $161/mo, below the bar

Bars are the Navigator Index score on a plain 0–100 scale; the tick is your bar, and the number beside each bar is the score itself. Monthly cost = cost per successful task × successful tasks, printed as a number. Failed runs are already priced in: cost per success divides everything spent by the tasks that passed. Definition

07Methodology & freshness

Every number, defined

14 Oct 2026data through
1,800runs
14entries
40tasks
v2.0Index
15 Oct 2026published

Demo data · not real results This board is built from development fixtures, not from a published cycle. Model names are placeholders and no number on it is a measurement.

Navigator Index
0–100. The weighted mean of 8 category scores; a category score is the mean task score (partial credit, 0–1) × 100 over that category’s active tasks and every repeat. Weights: Coding 15% · Data analysis 15% · Extraction 10% · Reasoning 15% · Research & agentic 15% · Summarization 10% · Support 10% · Writing 10%. Research & agentic: For now these are single-call multi-step tasks: one model call returns plan → steps → deliverable, checked step by step. Tool-using, browsing and long agentic tasks are deferred.
Intervals
95%, task-clustered: a bootstrap over tasks (stratified by category for the Index), never naive per-run intervals. Every score on this page shows its interval as a whisker or a range.
Ranks are ranges
Every pair of models is compared on the same tasks (a paired test of the Index difference, with family-wise error control, so the chance of any false “win” on the board is at most 5%). A rank is the range the data allows: best = 1 + models significantly better, worst = N − models significantly worse. “2–4” means we can't statistically separate it from the models in that range; a single number means it is pinned down.
Dominance (kill zone)
A model dominates another only when it is significantly better on the Index (the same paired test) and has a lower cost per successful task. A point-estimate gap inside the noise is never a kill. Reference rows are never counted.
Cost per successful task
Everything measured on candidate calls (OpenRouter's reported amounts) divided by the tasks passed, where a pass is a score at or above the task's threshold. Judge cost is excluded. Blank when nothing passed.
Latency
Median wall-clock latency per run, as measured on the pinned provider.
Movement
Δ compares with the previous published cycle at the same Index version only. Without one, the entry shows NEW or “first cycle”; movement is never estimated.
Commoditization curve
One point per published cycle: the cheapest cost per successful task among models at Index 70 or better. With fewer than three cycles it reads “tracking since”.
Frontier and quadrants
The frontier joins models that no other model beats on both Index and cost per success. Quadrants split at the quality bar (Index 70) and at the median plotted cost per success; they are reading aids, not verdicts.
Full methodologyChangelogReleasesCorrectionsIndependenceWhat changed in October 2026 (demo)Published comparisons (classic view)Presenter mode
DownloadsBoard API · available once a cycle is publishedJSON · available once a cycle is publishedCSV · available once a cycle is published