Showing examples

Does few shot prompting help?

Give one good example of the answer you want. A second or third adds very little.

Why it matters

One example pins the shape of the answer. Writing more of them costs you time and buys almost nothing.

How to use it

Paste one example in the format you need. Add more only for formats that are tricky to describe in words. This helped a little on 9 of 10 models tested, under our bar for a clear win.

Holds

None against three examples Pass rate up 18 to 19 points on all three models. Those results are tested via API. Also measured in Claude Code and Codex, reported separately on this page and never averaged with this.

None against one example Pass rate up 16 to 19 points on all three models. Those results are tested via API. Also measured in Claude Code and Codex, reported separately on this page and never averaged with this.

None against three examples

tested via API · Helped a little, under our bar on Claude Haiku 4.5, tested via API · Helped a little, under our bar on GPT-5 mini, tested via API · Helped a little, under our bar on Gemini 3.1 Flash Lite, tested via API

tested in Claude Code · Helped a little, under our bar on Claude Fable 5, tested in Claude Code · Helped a little, under our bar on Claude Opus 5, tested in Claude Code · Helped a little, under our bar on Claude Haiku 4.5, tested in Claude Code · Helped a little, under our bar on Claude Sonnet 5, tested in Claude Code · Helped a little, under our bar on Claude Fable 5.1, tested in Claude Code

tested in Codex · Holds on GPT-5.4 mini, Codex run on GPT-5.4 mini, tested in Codex · Helped a little, under our bar on GPT-5.6 Luna, Codex run on GPT-5.6 Luna, tested in Codex · Helped a little, under our bar on GPT-5.6 Terra, Codex run on GPT-5.6 Terra, tested in Codex

None against one example

tested via API · Helped a little, under our bar on Claude Haiku 4.5, tested via API · Helped a little, under our bar on GPT-5 mini, tested via API · Helped a little, under our bar on Gemini 3.1 Flash Lite, tested via API

tested in Claude Code · Helped a little, under our bar on Claude Fable 5, tested in Claude Code · Helped a little, under our bar on Claude Opus 5, tested in Claude Code · Helped a little, under our bar on Claude Haiku 4.5, tested in Claude Code · Helped a little, under our bar on Claude Sonnet 5, tested in Claude Code · Helped a little, under our bar on Claude Fable 5.1, tested in Claude Code

tested in Codex · Holds on GPT-5.4 mini, Codex run on GPT-5.4 mini, tested in Codex · Helped a little, under our bar on GPT-5.6 Luna, Codex run on GPT-5.6 Luna, tested in Codex · Helped a little, under our bar on GPT-5.6 Terra, Codex run on GPT-5.6 Terra, tested in Codex

The original set, retired from the table

tested in Claude Code · Could not measure on Claude Fable 5, tested in Claude Code · Could not measure on Claude Opus 5, tested in Claude Code · Could not measure on Claude Haiku 4.5, tested in Claude Code · Could not measure on Claude Sonnet 5, tested in Claude Code · Could not measure on Claude Fable 5.1, tested in Claude Code

tested in Codex · Could not measure on GPT-5.4 mini, Codex run on GPT-5.4 mini, tested in Codex · Could not measure on GPT-5.6 Luna, Codex run on GPT-5.6 Luna, tested in Codex · Could not measure on GPT-5.6 Terra, Codex run on GPT-5.6 Terra, tested in Codex

What the marks mean

  • Helped a little, under our bar
See the numbers per model

Without to with, per model

Via API tested via API

Claude Haiku 4.5 None against three examples

0.814
1.000

No measured effect

Gemini 3.1 Flash Lite None against three examples

0.800
0.986

No measured effect

GPT-5 mini None against three examples

0.794
0.977

No measured effect

Claude Haiku 4.5 None against one example

0.814
0.971

No measured effect

GPT-5 mini None against one example

0.783
0.969

No measured effect

Gemini 3.1 Flash Lite None against one example

0.800
0.957

No measured effect

deterministic pass rate, 0 to 1

In Claude Code tested in Claude Code

Claude Fable 5 None against three examples

0.820
1.000

No measured effect

Claude Fable 5.1 None against three examples

0.814
1.000

No measured effect

Claude Fable 5 The original set, retired from the table

1.000 without and with

Could not measure

Claude Fable 5.1 The original set, retired from the table

1.000 without and with

Could not measure

Claude Fable 5.1 None against one example

0.817
1.000

No measured effect

Claude Fable 5 None against one example

0.823
0.997

No measured effect

Claude Haiku 4.5 The original set, retired from the table

1.000 without and with

Could not measure

Claude Opus 5 The original set, retired from the table

1.000 without and with

Could not measure

Claude Sonnet 5 None against three examples

0.817
1.000

No measured effect

Claude Sonnet 5 The original set, retired from the table

1.000 without and with

Could not measure

Claude Opus 5 None against three examples

0.814
0.997

No measured effect

Claude Haiku 4.5 None against three examples

0.814
0.989

No measured effect

Claude Opus 5 None against one example

0.814
0.980

No measured effect

Claude Sonnet 5 None against one example

0.814
0.977

No measured effect

Claude Haiku 4.5 None against one example

0.814
0.943

No measured effect

deterministic pass rate, 0 to 1

A picture of the per-model numbers, drawn from the same results. The tables are the source. Grey is the score without. Colour is the score with. The bracket shows how much the difference could move if we ran it again. Rows are ordered by the score with. A model measured at two versions keeps its versions next to each other. The two test methods are reported separately and never averaged. Where the two scores are the same, the number is printed once at the end of the pair. A bracket wider than the axis is drawn to the edge with its cap omitted; the table gives its bounds.
Show per-task detail

Every item, without and with, per model

Via API tested via API

C04b-zero-vs-one

  • GPT-5 mini None against one example
  • Gemini 3.1 Flash Lite None against one example
  • Claude Haiku 4.5 None against one example

deterministic pass rate, 0 to 1

Hover or focus an item to read its task and every model’s two values.

Task-paired: these item means are what this cell's interval is built from.

The item means behind this plot
Per-item arm means. Without is the unaided arm, with is the treated arm.
ItemGPT-5 mini None against one example Gemini 3.1 Flash Lite None against one example Claude Haiku 4.5 None against one example
Task without with without with without with
t04 0.629 0.971 0.714 1.000 0.714 0.857
t02 0.714 0.857 0.714 0.857 0.714 1.000
t10 0.714 0.857 0.714 0.857 0.714 0.857
t01 0.771 1.000 0.714 0.857 0.857 1.000
t08 0.714 1.000 0.857 1.000 0.857 1.000
t03 0.857 1.000 0.857 1.000 0.857 1.000
t05 0.857 1.000 0.857 1.000 0.857 1.000
t06 0.857 1.000 0.857 1.000 0.857 1.000
t07 0.857 1.000 0.857 1.000 0.857 1.000
t09 0.857 1.000 0.857 1.000 0.857 1.000

C04b-zero-vs-three

  • GPT-5 mini None against three examples
  • Gemini 3.1 Flash Lite None against three examples
  • Claude Haiku 4.5 None against three examples

deterministic pass rate, 0 to 1

Hover or focus an item to read its task and every model’s two values.

Task-paired: these item means are what this cell's interval is built from.

The item means behind this plot
Per-item arm means. Without is the unaided arm, with is the treated arm.
ItemGPT-5 mini None against three examples Gemini 3.1 Flash Lite None against three examples Claude Haiku 4.5 None against three examples
Task without with without with without with
t04 0.657 0.914 0.714 1.000 0.714 1.000
t02 0.714 1.000 0.714 1.000 0.714 1.000
t10 0.714 0.857 0.714 1.000 0.714 1.000
t01 0.857 1.000 0.714 0.857 0.857 1.000
t08 0.714 1.000 0.857 1.000 0.857 1.000
t03 0.857 1.000 0.857 1.000 0.857 1.000
t05 0.857 1.000 0.857 1.000 0.857 1.000
t06 0.857 1.000 0.857 1.000 0.857 1.000
t07 0.857 1.000 0.857 1.000 0.857 1.000
t09 0.857 1.000 0.857 1.000 0.857 1.000

In Claude Code tested in Claude Code

C04-three-shot-format

  • Claude Fable 5 The original set, retired from the table
  • Claude Fable 5.1 The original set, retired from the table
  • Claude Haiku 4.5 The original set, retired from the table
  • Claude Opus 5 The original set, retired from the table
  • Claude Sonnet 5 The original set, retired from the table

deterministic pass rate, 0 to 1

Hover or focus an item to read its task and every model’s two values.

Pooled: the interval is taken over the records, not these means. These are those same records collapsed per item, so where an item carries more passes than another the lines need not average to the figure the table prints.

The item means behind this plot
Per-item arm means. Without is the unaided arm, with is the treated arm.
ItemClaude Fable 5 The original set, retired from the table Claude Fable 5.1 The original set, retired from the table Claude Haiku 4.5 The original set, retired from the table Claude Opus 5 The original set, retired from the table Claude Sonnet 5 The original set, retired from the table
Task without with without with without with without with without with
t01 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
t02 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
t03 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
t04 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
t05 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
t06 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
t07 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
t08 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
t09 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
t10 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

C04b-zero-vs-one

  • Claude Haiku 4.5 None against one example
  • Claude Opus 5 None against one example
  • Claude Sonnet 5 None against one example
  • Claude Fable 5.1 None against one example
  • Claude Fable 5 None against one example

deterministic pass rate, 0 to 1

Hover or focus an item to read its task and every model’s two values.

Task-paired: these item means are what this cell's interval is built from.

The item means behind this plot
Per-item arm means. Without is the unaided arm, with is the treated arm.
ItemClaude Haiku 4.5 None against one example Claude Opus 5 None against one example Claude Sonnet 5 None against one example Claude Fable 5.1 None against one example Claude Fable 5 None against one example
Task without with without with without with without with without with
t02 0.714 0.971 0.714 1.000 0.714 0.943 0.714 1.000 0.714 1.000
t10 0.714 0.714 0.714 0.914 0.714 0.857 0.714 1.000 0.714 0.971
t04 0.714 0.857 0.714 0.886 0.714 1.000 0.743 1.000 0.800 1.000
t01 0.857 1.000 0.857 1.000 0.857 0.971 0.857 1.000 0.857 1.000
t03 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000
t05 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000
t06 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000
t07 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000
t08 0.857 0.886 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000
t09 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000

C04b-zero-vs-three

  • Claude Fable 5.1 None against three examples
  • Claude Haiku 4.5 None against three examples
  • Claude Opus 5 None against three examples
  • Claude Sonnet 5 None against three examples
  • Claude Fable 5 None against three examples

deterministic pass rate, 0 to 1

Hover or focus an item to read its task and every model’s two values.

Task-paired: these item means are what this cell's interval is built from.

The item means behind this plot
Per-item arm means. Without is the unaided arm, with is the treated arm.
ItemClaude Fable 5.1 None against three examples Claude Haiku 4.5 None against three examples Claude Opus 5 None against three examples Claude Sonnet 5 None against three examples Claude Fable 5 None against three examples
Task without with without with without with without with without with
t02 0.714 1.000 0.714 1.000 0.714 1.000 0.714 1.000 0.714 1.000
t10 0.714 1.000 0.714 1.000 0.714 1.000 0.714 1.000 0.714 1.000
t04 0.714 1.000 0.714 1.000 0.714 0.971 0.743 1.000 0.771 1.000
t01 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000
t03 0.857 1.000 0.857 0.886 0.857 1.000 0.857 1.000 0.857 1.000
t05 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000
t06 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000
t07 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000
t08 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000
t09 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000 0.857 1.000
A picture of the per-item means behind the per-model numbers. The tables are the source. Each model draws two lines over the same task set: a dashed line through its unaided scores and a solid line through its treated ones. Items are ordered by the mean unaided score across the models that measured them, lowest first. The two instruments are reported separately and never averaged.

Result

One worked example raised the pass rate by 16 to 19 points on all three models tested via API, and showed no measured effect on all five models tested in Claude Code, and held on GPT-5.4 mini via the Codex run on GPT-5.4 mini, and showed no measured effect on GPT-5.6 Luna via the Codex run on GPT-5.6 Luna, and showed no measured effect on GPT-5.6 Terra via the Codex run on GPT-5.6 Terra. Measured 2026-08-13.

Show the per-model numbers

Claim tested: One worked example improves format compliance over no examples, on the same task.

Helped a little on Claude Haiku 4.5, GPT-5 mini and Gemini 3.1 Flash Lite, under our bar.

Circulates in practitioner communities. Tested because it circulates, not because it is endorsed.

This tip is OpenAddict's plain-language read of the measured result. The measurement below is the evidence, and it is what the reading has to answer to.

Ledger idC04b-zero-vs-one

Clarification, . Until 2026-08-31 this figure read 5, which was the pass count rather than a number of calls. What changed, and why it is filed once.

What was tested

This claim circulates in practitioner communities as advice about how to write prompts. That it circulates is an input to what gets tested here. It is a reason to test the claim, and it is not evidence for or against it. The result below is the evidence, and it is the only thing on this page that carries weight.

The comparison is paired. Two prompts differ in one respect, the manipulated variable, and are otherwise identical by construction. Nothing here supports a causal reading beyond that pairing.

Per-model numbers

Per-model results. Means are over valid scored records only. Invalid records are excluded from every denominator and counted in coverage.
MeasureClaude Haiku 4.5GPT-5 miniGemini 3.1 Flash LiteClaude Haiku 4.5GPT-5 miniGemini 3.1 Flash Lite
Control arm0.814n 100.794n 100.800n 100.814n 100.783n 100.800n 10
Treatment arm1.000n 100.977n 100.986n 100.971n 100.969n 100.957n 10
Delta+0.186+0.183+0.186+0.157+0.186+0.157
Interval, 95 percent0.143 to 0.2280.143 to 0.2230.143 to 0.2280.129 to 0.1850.140 to 0.2320.129 to 0.185
OrbitIn free drift100 of 100 recordsIn free drift100 of 100 recordsIn free drift100 of 100 recordsIn free drift100 of 100 recordsIn free drift100 of 100 recordsIn free drift100 of 100 records

Orbit is assigned by the frozen status_v1 rule. On this scale, deterministic pass rate, 0 to 1, the pass threshold is +0.20 and the failure floor is -0.20, each requiring an interval that excludes zero.

In Claude Code

These cells are tested in Claude Code, on a subscription path with no API key. They are a second instrument and are never averaged with the figures above, which are tested via API. What that means, and how it was calibrated.

Single-turn cells, replayed from the committed claims on the Claude Code CLI.
ModelControlTreatmentDeltaIntervalPairsOrbit
Claude Fable 51.00001.0000+0.0000.000 to 0.00050Unobservable
Claude Opus 51.00001.0000+0.0000.000 to 0.00050Unobservable
Claude Fable 50.82290.9971+0.1740.140 to 0.20810In free drift
Claude Opus 50.81430.9800+0.1660.137 to 0.19410In free drift
Claude Fable 50.82001.0000+0.1800.142 to 0.21810In free drift
Claude Opus 50.81430.9971+0.1830.143 to 0.22310In free drift
Claude Haiku 4.51.00001.0000+0.0000.000 to 0.00050Unobservable
Claude Sonnet 51.00001.0000+0.0000.000 to 0.00050Unobservable
Claude Haiku 4.50.81430.9429+0.1290.085 to 0.17210In free drift
Claude Sonnet 50.81430.9771+0.1630.130 to 0.19510In free drift
Claude Haiku 4.50.81430.9886+0.1740.122 to 0.22710In free drift
Claude Sonnet 50.81711.0000+0.1830.143 to 0.22310In free drift
Claude Fable 5.11.00001.0000+0.0000.000 to 0.00050Unobservable
Claude Fable 5.10.81711.0000+0.1830.143 to 0.22310In free drift
Claude Fable 5.10.81431.0000+0.1860.143 to 0.22810In free drift

First attempt (retired)

An earlier task set tested the same question and was withdrawn: every model scored full marks with and without it, so the set could not separate the arms; a hardened set replaced it. The rows below are that first attempt, and they are kept because a set that measured nothing is evidence about the instrument rather than about the technique.

Per-model results. Means are over valid scored records only. Invalid records are excluded from every denominator and counted in coverage.
MeasureClaude Haiku 4.5GPT-5 miniGemini 3.1 Flash Lite
Control arm1.000n 501.000n 501.000n 50
Treatment arm1.000n 501.000n 501.000n 50
Delta+0.000+0.000+0.000
Interval, 95 percent0.000 to 0.0000.000 to 0.0000.000 to 0.000
OrbitUnobservable100 of 100 recordsUnobservable100 of 100 recordsUnobservable100 of 100 records

Sets are now probed before they are funded, against a control-arm band that a saturated set cannot clear. How task sets are calibrated.

Method for this claim

Task set
10 messy multi-sentence stock records reduced to one pipe-separated line of seven fields. Hardened a FOURTH time after calibration. v3 passed at 0.886 but all of its headroom sat on one field: the control wrote PT401 where the key wanted PT-401, because the instruction said uppercase-no-spaces and never mentioned the hyphen. The pair would have measured one hyphen rather than convention-pinning. v4 removes four MORE statements from the instruction, so the convention sites are: SKU punctuation and casing, date format, flag separator, and flag ordering. Each is unstated, each is legitimately ambiguous, and each is pinned by the examples. Headroom now distributes across fields.
Runs per model per arm
5 passes x 10 items
Scoring
Deterministic, via scoreFieldwiseLine. A committed function scores each answer with no model in the loop.
Pass criterion as written for the pilot
Treatment mean fieldwise score exceeds control by at least 0.10 on the same 10 records.

The published verdict comes from status_v1, not from the pass criterion above. The criterion is recorded because it is what the claim was registered with before the run.

Model versions, as recorded

Read from the run records, not from configuration.
ModelVersion string returnedDelta on this claimInterval
Claude Haiku 4.5claude-haiku-4-5-20251001+0.1860.143 to 0.228
GPT-5 minigpt-5-mini-2025-08-07+0.1830.143 to 0.223
Gemini 3.1 Flash Litegemini-3.1-flash-lite+0.1860.143 to 0.228
Claude Haiku 4.5claude-haiku-4-5-20251001+0.1570.129 to 0.185
GPT-5 minigpt-5-mini-2025-08-07+0.1860.140 to 0.232
Gemini 3.1 Flash Litegemini-3.1-flash-lite+0.1570.129 to 0.185

Reading across models

Sampling was not held constant across vendors, so comparing one model column against another compares two settings as well as two models.

GPT-5 mini rejected the fixed sampling setting and ran at its own default on all 1,040 of its calls. The other models ran at temperature 0.

The tip above is editorial. Every figure inside the measurement is computed at build time from the committed pilot records by the status_v1 rule, and none of it is written by hand.