Splitting big tasks

Does prompt chaining work?

Break a big task into a few prompts, one step at a time.

Why it matters

A task with several parts gives a model several chances to drop one. Doing the parts in turn keeps each one in view.

How to use it

Ask for the first step, then feed its answer into the next prompt. Best on tasks with several parts. Check that it helps on the model you use.

Holds

Pass rate up 10 to 25 points on all three models. Those results are tested via API. Also measured in Claude Code, reported separately on this page and never averaged with this.

tested via API · Holds on Claude Haiku 4.5, tested via API · Holds on GPT-5 mini, tested via API · No measured effect on Gemini 3.1 Flash Lite, tested via API

tested in Claude Code · Holds on Claude Haiku 4.5, multi-turn, tested in Claude Code

What the marks mean

  • Holds
  • No measured effect
See the numbers per model

Without to with, per model

Via API tested via API

Claude Haiku 4.5

0.767
1.000

Holds

Gemini 3.1 Flash Lite

0.900
1.000

No measured effect

GPT-5 mini

0.753
1.000

Holds

deterministic pass rate, 0 to 1

In Claude Code tested in Claude Code

Claude Haiku 4.5 multi-turn

0.687
1.000

Holds

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. A bracket wider than the axis is drawn to the edge with its cap omitted; the table gives its bounds.
Show per-task detail

Without to with, per model

Via API tested via API

GPT-5 mini

Holds

Claude Haiku 4.5

Holds

Gemini 3.1 Flash Lite

No measured effect

deterministic pass rate, 0 to 1

In Claude Code tested in Claude Code

Claude Haiku 4.5 multi-turn

Holds

deterministic pass rate, 0 to 1

A picture of the per-model numbers, drawn from the same results. The tables are the source. Each row runs from the score without to the score with. The thin bar beneath it is the bracket, and it shows how much the difference could move if we ran it again. The two test methods are reported separately and never averaged. A bracket wider than the axis is drawn to the edge with its end cap omitted; the table gives its bounds.

The Claude Code readings on this page are the multi-turn run, which committed its comparison as arm means and an interval rather than as per-item values. A profile of the items behind that interval cannot be drawn from what it commits, so this page keeps the without-to-with chart.

Result

Breaking a complex task into separate sequential prompts raised the pass rate by 10 to 25 points on all three models tested via API, and held on Claude Haiku 4.5 via the multi-turn run in Claude Code. Measured 2026-08-13.

Show the per-model numbers

Claim tested: Breaking a complex task into separate sequential prompts beats one combined prompt.

Holds on Claude Haiku 4.5 and GPT-5 mini. No measured effect on Gemini 3.1 Flash Lite.

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 idC14-prompt-chaining

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 Lite
Control arm0.767n 100.753n 100.900n 10
Treatment arm1.000n 101.000n 101.000n 10
Delta+0.233+0.247+0.100
Interval, 95 percent0.134 to 0.3330.162 to 0.3320.000 to 0.200
OrbitStable100 of 100 recordsStable100 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.

Multi-turn cells. One model, and the interval beside its API counterpart below.
ModelControlTreatmentDeltaIntervalPairsOrbit
Claude Haiku 4.50.68671.0000+0.3130.176 to 0.45110Stable

The multi-turn run recorded intervals and a continuity comparison rather than a verdict. The Orbit above is assigned from those committed figures by the same status_v1 rule used everywhere else on this site; no threshold or interval is recomputed.

Multi-turn panel. Three models on one instrument, replayed from the same committed prompts.
ModelControlTreatmentDeltaIntervalPairsOrbit
Claude Opus 51.00001.0000+0.0000.000 to 0.00010Unobservable
Claude Sonnet 50.99331.0000+0.007-0.006 to 0.02010In free drift
Claude Fable 5.11.00001.0000+0.0000.000 to 0.00010Unobservable

These cells have no API counterpart. The wave 2 API records cover this claim on Claude Haiku 4.5 and on no other model, so nothing above is compared against an API interval and no continuity reading is taken from it: an overlap against a different model’s API cell would read a difference between models as a difference between instruments. The panel holds the instrument fixed and varies the model. The transport verdict is unchanged.

No row here says one model is better than another, and none attributes a difference to a cause: no model on this panel was run twice, so a cell that differs and a re-run of the same model are not distinguishable by this design. The Orbit is assigned from the committed figures by the same status_v1 rule used everywhere else on this site; no threshold or interval is recomputed.

Beside the panel, never pooled with it: an earlier reading on the same instrument.
ModelControlTreatmentDeltaIntervalPairsOrbit
Claude Haiku 4.5 (not on the panel)0.68671.0000+0.3130.176 to 0.45110Stable
Claude Fable 5.1 (earlier pass)1.00001.0000+0.0000.000 to 0.00010Unobservable

The earlier pass is the same model measured on the same instrument under an older gate, before records carried a schema version, an invocation id or the served-model rule. It is a second reading, not an average: two passes of one model are two numbers, and averaging them would hide whichever of the two moved, which is the only thing a repeat pass is good for.

The Claude Code interval on this claim overlaps its API interval. Across the multi-turn run the pre-registered comparison returned "continuous with panel v1, 3 of 3 cells", and that statement covers the cells it measured and is not extrapolated to any cell that was not run.

Method for this claim

Task set
10 three-part tasks of the extract-then-compute-then-report class. The control arm asks for all three parts in one prompt; the treatment arm asks for them in three turns, each seeing the previous answer verbatim. Scored only on the final output, against committed literal keys: the computed figures must appear and the stated shape must hold.
Runs per model per arm
5 passes x 10 items
Scoring
Deterministic, via scoreRequirementKeys. A committed function scores each answer with no model in the loop.
Pass criterion as written for the pilot
Treatment mean requirement satisfaction exceeds control by at least 0.10 on the same 10 tasks.

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+0.2330.134 to 0.333
GPT-5 minigpt-5-mini-2025-08-07+0.2470.162 to 0.332
Gemini 3.1 Flash Litegemini-3.1-flash-lite+0.1000.000 to 0.200

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.