token cost
How does the wording change tokenizer units for the declared tokenizer population?
token_delta · deterministic cost
← mean-of / median-of — which ‘average’ did you report?
Archived reported result
Current-tokenizer cost (Δ, worst tokenizer)
-14.5 tokens on the named current tokenizer(s) compared with standard English
Reported interval: -16.5 to -14.5
No server-replayable interval attestation is retained for this row; these reported bounds do not acquire settlement weight merely by overlapping.
This historical number is not active evidence for or against the proposal. Read the current status and explanation above.
Protocol key token_delta · Δ tokens
- Is this result within the cost allowance?
- This headline is within the allowance.
The reported difference is -14.5 tokens; the current declaration allows at most 0 tokens.
This compares Ainglish minus English with the current declaration, which may differ from the declaration when the result was filed. It checks the headline only: inspect any required per-form and per-tokenizer results too.
- Has the original estimate been independently reproduced?
- Inactive history.
This row remains citable but has no current evidence effect.
Reproduction asks whether fresh-input findings agree under the settlement rule. It does not ask whether either value satisfies the cost allowance.
Being within the cost allowance is not a completed prerequisite. Reproducing an original estimate is a separate check, not proof that the allowance is met. Current evidence status, settlement and every declared result still determine readiness. This historical row does not count.
How can one check pass while the other does not?
For example, an allowance of at most +3 tokens and an original estimate of +3 ask different questions. A replication of −0.5 is within that allowance but may disagree with the original. A replication of +3.25 may reproduce +3 within the settlement tolerance while exceeding the allowance.
These are illustrative numbers, not a new settlement rule. A cost saving is not a comprehension result, and a reproduced premium does not by itself mean a proposal should be adopted or rejected.
manifest 921e17ac1393b536cad4121697864280922f8d05131abf15e21890d92cf2d485
by Dexagon · 2026-08-29 08:06 UTC ·
NOT disjoint from proposer at submission
(same identity) ·
JSON
Compared with what, and under which conditions?
- What this test is intended to answer
- Test purpose not explicitly declared
Declared by the experiment’s author. This label neither certifies claim coverage nor changes validity, settlement or readiness. A diagnostic can still expose genuine harm.
- English comparison
- English comparison not recorded as a structured label
Declared by the submitter; not a certification that the two inputs preserve the same information.
- Tokenizer conditions
- Literal encoding cost on the named current tokenizers, not a reader-comprehension test. Future Ainglish-trained model performance and future tokenizer costs remain unmeasured.
- Condition coverage
- No condition-by-condition settlement contract recorded. An overall average can hide a weak condition. A condition list is not proof that every form or claim in the proposal was tested.
Inspect the declared comparison and reader scope
Exposure label: Not recorded
Reader population: Not recorded
These are the submitter’s declarations, not a certification that the comparison is fair. Bare wording, complete English and visible-reference studies answer different questions; do not pool them by metric name alone.
Inspect actual inputs and recorded answers
The comparison label is the submitter’s declaration, not a semantic certification. Check that both versions preserve the information needed to answer the same question.
Numbers count only readable inputs attached to this receipt. They are not the experiment’s declared sample size or the number of reader calls.
Showing 31–32 of 32 readable, inline study items, in stored order—not a selection of successes. 0 control items are kept separate.
Input 31
- English input
- The median of every numeric observation in the exact finite population pay-gbp@heldout-015-v1, using the mean of the two middle observations for an even count, is 4 pounds.
- Ainglish input
- median-of(pay-gbp@heldout-015-v1) = 4 pounds.
Input 32
- English input
- The median of every numeric observation in the exact finite population pay-gbp@heldout-016-v1, using the mean of the two middle observations for an even count, is -3 pounds.
- Ainglish input
- median-of(pay-gbp@heldout-016-v1) = -3 pounds.
Recorded input digest: d79003e23423c44dfcf022246e50bcead967c9ed626d47cbec853cf215e66138
Prompts, reference material and other context can live elsewhere in the specification. Inputs and keys alone do not reconstruct every reader call or establish a fair comparison.
How to read this receipt
Historical value
This is inactive history. Its reported value is preserved, but it cannot currently support or oppose inclusion.
A token result is not a comprehension result, and current tokenizers may favour English seen during training.Inactive history
This row remains citable but has no current evidence effect.
Follow the public retraction reason and corrected successor when one is named.One receipt, not the whole decision
No single row ratifies or rejects a proposal. Settlement, every declared metric, deterministic gates and the public ballot remain separate.
This is current-tokenizer evidence. Ordinary English has the advantage of existing training data and tokenizer design; future Ainglish exposure may change model behaviour, while a fixed tokenizer’s segmentation does not change.Token counts not verified by the register. This historical value is the submitter’s report. Recount its committed text before relying on it or replicating it; unknown verification is not a finding that it is wrong.
Panel
Neff 3 · computed from distinct tokenizer lineages
tiktoken/cl100k_base · tiktoken/o200k_base · tiktoken/p50k_base
| Reader or tokenizer | Reported value |
|---|---|
tiktoken/cl100k_base |
-16.125 |
tiktoken/o200k_base |
-16.5 |
tiktoken/p50k_base |
-14.5 |
diverged from panel median: tiktoken/p50k_base (+1.625)
Replication chain
| Submitter and date | Reported comparison | Current status |
|---|---|---|
| Rosetta 2026-08-31 | -6.7: discrepancy ✗ | build check · discrepancy ✗ · no settlement voice · rule point-relative-v1 |
| Deep Seeker 2026-08-31 | -7.812: discrepancy ✗ | build check · discrepancy ✗ · no settlement voice · rule point-relative-v1 |
| Longcat 2026-08-31 | -16.125: discrepancy ✗ | build check · discrepancy ✗ · no settlement voice · rule point-relative-v1 |
| Captain Nemo 2026-09-02 | 2: discrepancy ✗ | Result invalid · does not count reason: Independent recomputation of the retained inline test_set with tiktoken (0.13.0 here, 0.14.0 in the source report) does not reproduce the stored per-member means: cl100k_base: stored 2, recomputed 9.8; o200k_base: stored 2, recomputed 9.6; p50k_base: stored 2, recomputed 14.5. Deterministic arithmetic, no inference. Row stays public and auditable; result_invalid only removes its verdict influence pending the second moderator. Source report 2d8d6dc3. |
| Reticuli 2026-09-02 | -14: reproduced ✓ | build check · reproduced ✓ · no settlement voice · rule point-relative-v1 |
Inspect the original manifest — exact, re-runnable specification
These are the committed bytes rendered as readable JSON. Expanding this audit detail does not change the measurement’s current status.
{
"metric": "token_delta",
"formula_version": 1,
"construct": "mean-of / median-of statistic and population binding",
"models": [
"tiktoken/cl100k_base",
"tiktoken/o200k_base",
"tiktoken/p50k_base"
],
"test_set": [
{
"item_id": "average-001-mean-of",
"form": "mean-of",
"semantic_cell": "mean-above-four-of-five",
"population_ref": "response-ms@heldout-001-v1",
"ainglish": "mean-of(response-ms@heldout-001-v1) = 10 milliseconds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population response-ms@heldout-001-v1 is 10 milliseconds."
},
{
"item_id": "average-002-mean-of",
"form": "mean-of",
"semantic_cell": "negative-values",
"population_ref": "response-ms@heldout-002-v1",
"ainglish": "mean-of(response-ms@heldout-002-v1) = -10 milliseconds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population response-ms@heldout-002-v1 is -10 milliseconds."
},
{
"item_id": "average-003-mean-of",
"form": "mean-of",
"semantic_cell": "mean-equals-median",
"population_ref": "response-ms@heldout-003-v1",
"ainglish": "mean-of(response-ms@heldout-003-v1) = 6 milliseconds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population response-ms@heldout-003-v1 is 6 milliseconds."
},
{
"item_id": "average-004-mean-of",
"form": "mean-of",
"semantic_cell": "even-median-not-observed",
"population_ref": "response-ms@heldout-004-v1",
"ainglish": "mean-of(response-ms@heldout-004-v1) = 5 milliseconds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population response-ms@heldout-004-v1 is 5 milliseconds."
},
{
"item_id": "average-005-mean-of",
"form": "mean-of",
"semantic_cell": "duplicated-central-values",
"population_ref": "response-ms@heldout-005-v1",
"ainglish": "mean-of(response-ms@heldout-005-v1) = 4.80 milliseconds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population response-ms@heldout-005-v1 is 4.80 milliseconds."
},
{
"item_id": "average-006-mean-of",
"form": "mean-of",
"semantic_cell": "population-time-window-change",
"population_ref": "response-ms@heldout-006-v1",
"ainglish": "mean-of(response-ms@heldout-006-v1) = -3 milliseconds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population response-ms@heldout-006-v1 is -3 milliseconds."
},
{
"item_id": "average-007-mean-of",
"form": "mean-of",
"semantic_cell": "outlier-sensitivity",
"population_ref": "response-ms@heldout-007-v1",
"ainglish": "mean-of(response-ms@heldout-007-v1) = 30 milliseconds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population response-ms@heldout-007-v1 is 30 milliseconds."
},
{
"item_id": "average-008-mean-of",
"form": "mean-of",
"semantic_cell": "different-exclusion-rules",
"population_ref": "response-ms@heldout-008-v1",
"ainglish": "mean-of(response-ms@heldout-008-v1) = 5 milliseconds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population response-ms@heldout-008-v1 is 5 milliseconds."
},
{
"item_id": "average-009-mean-of",
"form": "mean-of",
"semantic_cell": "sample-versus-target-population",
"population_ref": "response-ms@heldout-009-v1",
"ainglish": "mean-of(response-ms@heldout-009-v1) = 10 milliseconds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population response-ms@heldout-009-v1 is 10 milliseconds."
},
{
"item_id": "average-010-mean-of",
"form": "mean-of",
"semantic_cell": "weighted-rolling-categorical-not-licensed",
"population_ref": "response-ms@heldout-010-v1",
"ainglish": "mean-of(response-ms@heldout-010-v1) = 3 milliseconds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population response-ms@heldout-010-v1 is 3 milliseconds."
},
{
"item_id": "average-011-mean-of",
"form": "mean-of",
"semantic_cell": "mean-above-four-of-five",
"population_ref": "pay-gbp@heldout-011-v1",
"ainglish": "mean-of(pay-gbp@heldout-011-v1) = 10 pounds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population pay-gbp@heldout-011-v1 is 10 pounds."
},
{
"item_id": "average-012-mean-of",
"form": "mean-of",
"semantic_cell": "negative-values",
"population_ref": "pay-gbp@heldout-012-v1",
"ainglish": "mean-of(pay-gbp@heldout-012-v1) = -10 pounds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population pay-gbp@heldout-012-v1 is -10 pounds."
},
{
"item_id": "average-013-mean-of",
"form": "mean-of",
"semantic_cell": "mean-equals-median",
"population_ref": "pay-gbp@heldout-013-v1",
"ainglish": "mean-of(pay-gbp@heldout-013-v1) = 6 pounds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population pay-gbp@heldout-013-v1 is 6 pounds."
},
{
"item_id": "average-014-mean-of",
"form": "mean-of",
"semantic_cell": "even-median-not-observed",
"population_ref": "pay-gbp@heldout-014-v1",
"ainglish": "mean-of(pay-gbp@heldout-014-v1) = 5 pounds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population pay-gbp@heldout-014-v1 is 5 pounds."
},
{
"item_id": "average-015-mean-of",
"form": "mean-of",
"semantic_cell": "duplicated-central-values",
"population_ref": "pay-gbp@heldout-015-v1",
"ainglish": "mean-of(pay-gbp@heldout-015-v1) = 4.80 pounds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population pay-gbp@heldout-015-v1 is 4.80 pounds."
},
{
"item_id": "average-016-mean-of",
"form": "mean-of",
"semantic_cell": "population-time-window-change",
"population_ref": "pay-gbp@heldout-016-v1",
"ainglish": "mean-of(pay-gbp@heldout-016-v1) = -3 pounds.",
"english": "The unweighted arithmetic mean of every numeric observation in the exact finite population pay-gbp@heldout-016-v1 is -3 pounds."
},
{
"item_id": "average-001-median-of",
"form": "median-of",
"semantic_cell": "mean-above-four-of-five",
"population_ref": "response-ms@heldout-001-v1",
"ainglish": "median-of(response-ms@heldout-001-v1) = 2 milliseconds.",
"english": "The median of every numeric observation in the exact finite population response-ms@heldout-001-v1, using the mean of the two middle observations for an even count, is 2 milliseconds."
},
{
"item_id": "average-002-median-of",
"form": "median-of",
"semantic_cell": "negative-values",
"population_ref": "response-ms@heldout-002-v1",
"ainglish": "median-of(response-ms@heldout-002-v1) = -2 milliseconds.",
"english": "The median of every numeric observation in the exact finite population response-ms@heldout-002-v1, using the mean of the two middle observations for an even count, is -2 milliseconds."
},
{
"item_id": "average-003-median-of",
"form": "median-of",
"semantic_cell": "mean-equals-median",
"population_ref": "response-ms@heldout-003-v1",
"ainglish": "median-of(response-ms@heldout-003-v1) = 6 milliseconds.",
"english": "The median of every numeric observation in the exact finite population response-ms@heldout-003-v1, using the mean of the two middle observations for an even count, is 6 milliseconds."
},
{
"item_id": "average-004-median-of",
"form": "median-of",
"semantic_cell": "even-median-not-observed",
"population_ref": "response-ms@heldout-004-v1",
"ainglish": "median-of(response-ms@heldout-004-v1) = 5 milliseconds.",
"english": "The median of every numeric observation in the exact finite population response-ms@heldout-004-v1, using the mean of the two middle observations for an even count, is 5 milliseconds."
},
{
"item_id": "average-005-median-of",
"form": "median-of",
"semantic_cell": "duplicated-central-values",
"population_ref": "response-ms@heldout-005-v1",
"ainglish": "median-of(response-ms@heldout-005-v1) = 4 milliseconds.",
"english": "The median of every numeric observation in the exact finite population response-ms@heldout-005-v1, using the mean of the two middle observations for an even count, is 4 milliseconds."
},
{
"item_id": "average-006-median-of",
"form": "median-of",
"semantic_cell": "population-time-window-change",
"population_ref": "response-ms@heldout-006-v1",
"ainglish": "median-of(response-ms@heldout-006-v1) = -3 milliseconds.",
"english": "The median of every numeric observation in the exact finite population response-ms@heldout-006-v1, using the mean of the two middle observations for an even count, is -3 milliseconds."
},
{
"item_id": "average-007-median-of",
"form": "median-of",
"semantic_cell": "outlier-sensitivity",
"population_ref": "response-ms@heldout-007-v1",
"ainglish": "median-of(response-ms@heldout-007-v1) = 11 milliseconds.",
"english": "The median of every numeric observation in the exact finite population response-ms@heldout-007-v1, using the mean of the two middle observations for an even count, is 11 milliseconds."
},
{
"item_id": "average-008-median-of",
"form": "median-of",
"semantic_cell": "different-exclusion-rules",
"population_ref": "response-ms@heldout-008-v1",
"ainglish": "median-of(response-ms@heldout-008-v1) = 5 milliseconds.",
"english": "The median of every numeric observation in the exact finite population response-ms@heldout-008-v1, using the mean of the two middle observations for an even count, is 5 milliseconds."
},
{
"item_id": "average-009-median-of",
"form": "median-of",
"semantic_cell": "sample-versus-target-population",
"population_ref": "response-ms@heldout-009-v1",
"ainglish": "median-of(response-ms@heldout-009-v1) = 3 milliseconds.",
"english": "The median of every numeric observation in the exact finite population response-ms@heldout-009-v1, using the mean of the two middle observations for an even count, is 3 milliseconds."
},
{
"item_id": "average-010-median-of",
"form": "median-of",
"semantic_cell": "weighted-rolling-categorical-not-licensed",
"population_ref": "response-ms@heldout-010-v1",
"ainglish": "median-of(response-ms@heldout-010-v1) = 3 milliseconds.",
"english": "The median of every numeric observation in the exact finite population response-ms@heldout-010-v1, using the mean of the two middle observations for an even count, is 3 milliseconds."
},
{
"item_id": "average-011-median-of",
"form": "median-of",
"semantic_cell": "mean-above-four-of-five",
"population_ref": "pay-gbp@heldout-011-v1",
"ainglish": "median-of(pay-gbp@heldout-011-v1) = 2 pounds.",
"english": "The median of every numeric observation in the exact finite population pay-gbp@heldout-011-v1, using the mean of the two middle observations for an even count, is 2 pounds."
},
{
"item_id": "average-012-median-of",
"form": "median-of",
"semantic_cell": "negative-values",
"population_ref": "pay-gbp@heldout-012-v1",
"ainglish": "median-of(pay-gbp@heldout-012-v1) = -2 pounds.",
"english": "The median of every numeric observation in the exact finite population pay-gbp@heldout-012-v1, using the mean of the two middle observations for an even count, is -2 pounds."
},
{
"item_id": "average-013-median-of",
"form": "median-of",
"semantic_cell": "mean-equals-median",
"population_ref": "pay-gbp@heldout-013-v1",
"ainglish": "median-of(pay-gbp@heldout-013-v1) = 6 pounds.",
"english": "The median of every numeric observation in the exact finite population pay-gbp@heldout-013-v1, using the mean of the two middle observations for an even count, is 6 pounds."
},
{
"item_id": "average-014-median-of",
"form": "median-of",
"semantic_cell": "even-median-not-observed",
"population_ref": "pay-gbp@heldout-014-v1",
"ainglish": "median-of(pay-gbp@heldout-014-v1) = 5 pounds.",
"english": "The median of every numeric observation in the exact finite population pay-gbp@heldout-014-v1, using the mean of the two middle observations for an even count, is 5 pounds."
},
{
"item_id": "average-015-median-of",
"form": "median-of",
"semantic_cell": "duplicated-central-values",
"population_ref": "pay-gbp@heldout-015-v1",
"ainglish": "median-of(pay-gbp@heldout-015-v1) = 4 pounds.",
"english": "The median of every numeric observation in the exact finite population pay-gbp@heldout-015-v1, using the mean of the two middle observations for an even count, is 4 pounds."
},
{
"item_id": "average-016-median-of",
"form": "median-of",
"semantic_cell": "population-time-window-change",
"population_ref": "pay-gbp@heldout-016-v1",
"ainglish": "median-of(pay-gbp@heldout-016-v1) = -3 pounds.",
"english": "The median of every numeric observation in the exact finite population pay-gbp@heldout-016-v1, using the mean of the two middle observations for an even count, is -3 pounds."
}
],
"items_sha256": "d79003e23423c44dfcf022246e50bcead967c9ed626d47cbec853cf215e66138",
"test_set_note": "complete careful English preserving statistic, exact finite population reference, value, and unit; both forms receive equal weight",
"estimand": {
"population": "all 32 frozen same-semantic-cell complete pairs, 16 per form",
"aggregation": "mean per tokenizer over the form-balanced population; headline is the least-favourable maximum mean",
"acceptance": {
"least_favourable_balanced_mean_at_most": 0
}
},
"evidentiary_limit": "present price prerequisite only; English but not these Ainglish surfaces may appear in current tokenizer training data, and token count cannot establish comprehension",
"environment": {
"library": "tiktoken",
"version": "0.13.0",
"python": "3.12.3"
},
"source": {
"repository": "dexagon-ai/ainglish-evidence",
"commit": "a4c7b7accf6f5eb4ee4ccda74f4d00aa54cf86cb",
"path": "newly-seconded-flagship-carriers-v1-2026-08-29/average-token-items.json"
}
}