Accuracy, checked

Twenty-five facts. Our answer with its source,beside the chatbot's answer, verbatim.

Teachers ask chatbots for element facts because it is fast. The trouble is that a chatbot answers from what it half-remembers, and it sounds exactly as confident when it is wrong. Every number elementapi puts in a quiz comes from a curated record with a primary-source citation. This page shows what that difference looks like on twenty-five ordinary questions.

elementapi

25 / 25 cited

Every answer links to the record and the source it came from.

ChatGPT's free-tier model

23 / 25 correct

GPT-5.6 Luna, as ChatGPT Free answers (no Think). 1 wrong, 1 partly right.

With reasoning turned on

GPT-5.6 Luna with reasoning on (Think): 24 of 25 correct, 1 wrong.
GPT-5.5 (flagship) with reasoning on: 23 of 25 correct, 1 partly, 1 wrong.

Collected 2026-10-13 through the OpenAI API: model gpt-5.6-luna, reasoning effort none. One request per question, the prompt alone, no system prompt, first response. Prompts are printed in full so you can rerun them.

Most abundant isotope

How we judged: Correct if the isotope matches and the percentage is within 0.3 percentage points of the cited value (isotopic-composition evaluations differ at the second decimal). Naming the wrong isotope is incorrect regardless of the percentage.

1. Tin · Z = 50
“What is the most abundant naturally occurring isotope of tin, and what percentage of natural tin is that isotope? Give the isotope and one percentage.”
elementapi
¹²⁰Sn, 32.58% of natural tin (± 0.09)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Sn.json rev. 2026-10-13.
gpt-5.6-luna
Correct
Tin-120 (¹²⁰Sn), approximately **32.6%** of natural tin.
GPT-5.6 Luna with reasoning on (Think)Correct
Tin-120 (¹²⁰Sn), approximately **32.6%** of natural tin.
GPT-5.5 (flagship) with reasoning onCorrect
The most abundant naturally occurring isotope of tin is **tin-120 (^120Sn)**, making up about **32.58%** of natural tin.
2. Xenon · Z = 54
“What is the most abundant naturally occurring isotope of xenon, and what percentage of natural xenon is that isotope? Give the isotope and one percentage.”
elementapi
¹³²Xe, 26.909% of natural xenon (± 0.055)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Xe.json rev. 2026-10-13.
gpt-5.6-luna
Correct
The most abundant naturally occurring xenon isotope is **xenon-132 (¹³²Xe)**, comprising approximately **26.9%** of natural xenon.
GPT-5.6 Luna with reasoning on (Think)Correct
Xenon-132 (^132Xe) is the most abundant naturally occurring isotope, comprising approximately **26.9%** of natural xenon.
GPT-5.5 (flagship) with reasoning onCorrect
Xenon-132 (**¹³²Xe**) — about **26.9%** of natural xenon.
3. Tellurium · Z = 52
“What is the most abundant naturally occurring isotope of tellurium, and what percentage of natural tellurium is that isotope? Give the isotope and one percentage.”
elementapi
¹³⁰Te, 34.08% of natural tellurium (± 0.62)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Te.json rev. 2026-10-13.
gpt-5.6-luna
Correct
The most abundant naturally occurring isotope of tellurium is **tellurium-130 (¹³⁰Te)**, comprising approximately **34.1%** of natural tellurium.
GPT-5.6 Luna with reasoning on (Think)Correct
Tellurium-130 (¹³⁰Te) — approximately **34.08%** of naturally occurring tellurium.
GPT-5.5 (flagship) with reasoning onCorrect
Tellurium-130 (**¹³⁰Te**) is the most abundant naturally occurring isotope, making up about **34.08%** of natural tellurium.
4. Molybdenum · Z = 42
“What is the most abundant naturally occurring isotope of molybdenum, and what percentage of natural molybdenum is that isotope? Give the isotope and one percentage.”
elementapi
⁹⁸Mo, 24.292% of natural molybdenum (± 0.08)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Mo.json rev. 2026-10-13.
gpt-5.6-luna
Correct
Molybdenum-98 (⁹⁸Mo), approximately **24.14%** of natural molybdenum.
Why: 24.14% is an older evaluation value; the current one is 24.292%. 0.15 points off, inside the tolerance.
GPT-5.6 Luna with reasoning on (Think)Correct
Molybdenum-98 (⁹⁸Mo) — **24.13%** of natural molybdenum.
Why: 24.13% is an older evaluation value; the current one is 24.292%. 0.16 points off, inside the tolerance.
GPT-5.5 (flagship) with reasoning onCorrect
Molybdenum-98 (⁹⁸Mo), 24.13% of natural molybdenum.
Why: 24.13% is an older evaluation value; the current one is 24.292%. 0.16 points off, inside the tolerance.
5. Palladium · Z = 46
“What is the most abundant naturally occurring isotope of palladium, and what percentage of natural palladium is that isotope? Give the isotope and one percentage.”
elementapi
¹⁰⁶Pd, 27.33% of natural palladium (± 0.03)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Pd.json rev. 2026-10-13.
gpt-5.6-luna
Correct
Palladium-106 (¹⁰⁶Pd) — approximately **27.33%** of natural palladium.
GPT-5.6 Luna with reasoning on (Think)Correct
**Palladium-106 (¹⁰⁶Pd)** — approximately **27.3%** of natural palladium.
GPT-5.5 (flagship) with reasoning onCorrect
The most abundant naturally occurring isotope of palladium is **palladium-106 (^106Pd)**, which makes up about **27.33%** of natural palladium.
6. Germanium · Z = 32
“What is the most abundant naturally occurring isotope of germanium, and what percentage of natural germanium is that isotope? Give the isotope and one percentage.”
elementapi
⁷⁴Ge, 36.52% of natural germanium (± 0.12)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Ge.json rev. 2026-10-13.
gpt-5.6-luna
Correct
Germanium-⁷⁴ (⁷⁴Ge), approximately **36.5%** of natural germanium.
GPT-5.6 Luna with reasoning on (Think)Correct
Germanium-74 (⁷⁴Ge) is the most abundant naturally occurring isotope, comprising approximately **36.7%** of natural germanium.
Why: 36.7% is an older evaluation value; the current one is 36.52%. 0.18 points off, inside the tolerance.
GPT-5.5 (flagship) with reasoning onCorrect
Germanium-74 (⁷⁴Ge), about **36.5%** of natural germanium.
7. Osmium · Z = 76
“What is the most abundant naturally occurring isotope of osmium, and what percentage of natural osmium is that isotope? Give the isotope and one percentage.”
elementapi
¹⁹²Os, 40.78% of natural osmium (± 0.32)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Os.json rev. 2026-10-13.
gpt-5.6-luna
Correct
Osmium-192 (¹⁹²Os), approximately **41.0%** of natural osmium.
Why: 41.0% is an older evaluation value; the current one is 40.78%. 0.22 points off, inside the tolerance.
GPT-5.6 Luna with reasoning on (Think)Correct
Osmium-192 (¹⁹²Os) — approximately **40.78%** of natural osmium.
GPT-5.5 (flagship) with reasoning onCorrect
The most abundant naturally occurring isotope of osmium is **osmium-192 (^192Os)**, which makes up about **40.78%** of natural osmium.
8. Dysprosium · Z = 66
“What is the most abundant naturally occurring isotope of dysprosium, and what percentage of natural dysprosium is that isotope? Give the isotope and one percentage.”
elementapi
¹⁶⁴Dy, 28.26% of natural dysprosium (± 0.054)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Dy.json rev. 2026-10-13.
gpt-5.6-luna
Correct
The most abundant naturally occurring isotope of dysprosium is **dysprosium-164 (¹⁶⁴Dy)**, comprising approximately **28.18%** of natural dysprosium.
GPT-5.6 Luna with reasoning on (Think)Correct
Dysprosium-164 (¹⁶⁴Dy) is the most abundant naturally occurring isotope, comprising approximately **28.18%** of natural dysprosium.
GPT-5.5 (flagship) with reasoning onCorrect
**Dysprosium-164 (¹⁶⁴Dy), about 28.26%** of natural dysprosium.
9. Platinum · Z = 78
“What is the most abundant naturally occurring isotope of platinum, and what percentage of natural platinum is that isotope? Give the isotope and one percentage.”
elementapi
¹⁹⁵Pt, 33.775% of natural platinum (± 0.24)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Pt.json rev. 2026-10-13.
gpt-5.6-luna
Correct
Platinum-195 (¹⁹⁵Pt), approximately **33.8%** of natural platinum.
GPT-5.6 Luna with reasoning on (Think)Correct
Platinum-195 (^195Pt) is the most abundant naturally occurring isotope, comprising approximately **33.83%** of natural platinum.
GPT-5.5 (flagship) with reasoning onCorrect
Platinum-195 (^195Pt), about 33.8% of natural platinum.

Electron configuration

How we judged: Correct if every subshell occupancy matches the cited ground-state configuration. Notation and subshell order are free; a different occupancy (for example 4d8 5s2 instead of 4d10) is incorrect.

10. Niobium · Z = 41
“What is the ground-state electron configuration of a neutral niobium atom? Give it in noble-gas shorthand.”
elementapi
[Kr] 4d4 5s1
Source: PubChem Periodic Table, National Center for Biotechnology Information (NCBI), U.S. National Library of Medicine. Record: Nb.json rev. 2026-10-13.
gpt-5.6-luna
Correct
The ground-state electron configuration of neutral niobium (Nb, atomic number 41) is: \[ \boxed{[\mathrm{Kr}]\,4d^4\,5s^1} \] Niobium is an exception to the expected filling order, which would suggest \([\mathrm{Kr}]\,4d^3\,5s^2\).
GPT-5.6 Luna with reasoning on (Think)Correct
The ground-state electron configuration of neutral niobium (Nb, \(Z=41\)) is: \[ \boxed{[\mathrm{Kr}]\,4d^{4}5s^{1}} \]
GPT-5.5 (flagship) with reasoning onCorrect
The ground-state electron configuration of neutral niobium (Nb, Z = 41) is: **[Kr] 4d⁴ 5s¹**
11. Ruthenium · Z = 44
“What is the ground-state electron configuration of a neutral ruthenium atom? Give it in noble-gas shorthand.”
elementapi
[Kr] 4d7 5s1
Source: PubChem Periodic Table, National Center for Biotechnology Information (NCBI), U.S. National Library of Medicine. Record: Ru.json rev. 2026-10-13.
gpt-5.6-luna
Correct
The ground-state electron configuration of neutral ruthenium (Ru, atomic number 44) is: **[Kr] 4d⁷ 5s¹**
GPT-5.6 Luna with reasoning on (Think)Correct
The ground-state electron configuration of a neutral ruthenium atom (Ru, \(Z=44\)) is: \[ \boxed{[\mathrm{Kr}]\,4d^{7}5s^{1}} \]
GPT-5.5 (flagship) with reasoning onCorrect
The ground-state electron configuration of neutral ruthenium (Ru) is: **[Kr] 4d⁷ 5s¹**
12. Palladium · Z = 46
“What is the ground-state electron configuration of a neutral palladium atom? Give it in noble-gas shorthand.”
elementapi
[Kr] 4d10
Source: PubChem Periodic Table, National Center for Biotechnology Information (NCBI), U.S. National Library of Medicine. Record: Pd.json rev. 2026-10-13.
gpt-5.6-luna
Correct
The ground-state electron configuration of a neutral palladium atom (Pd, atomic number 46) is: **[Kr] 4d¹⁰** Palladium is an exception to the expected filling pattern because its 5s orbital is empty.
GPT-5.6 Luna with reasoning on (Think)Correct
The ground-state electron configuration of neutral palladium (Pd, \(Z=46\)) is: \[ \boxed{[\mathrm{Kr}]\,4d^{10}} \] (\(5s^0\) is implied.)
GPT-5.5 (flagship) with reasoning onCorrect
The ground-state electron configuration of neutral palladium (Pd) is: \[ \boxed{[\mathrm{Kr}]\,4d^{10}} \] Palladium is an exception to the usual filling pattern; its \(5s\) subshell is empty in the ground state.
13. Platinum · Z = 78
“What is the ground-state electron configuration of a neutral platinum atom? Give it in noble-gas shorthand.”
elementapi
[Xe] 4f14 5d9 6s1
Source: PubChem Periodic Table, National Center for Biotechnology Information (NCBI), U.S. National Library of Medicine. Record: Pt.json rev. 2026-10-13.
gpt-5.6-luna
Correct
The ground-state electron configuration of a neutral platinum atom (Pt, atomic number 78) is: \[ \boxed{[\mathrm{Xe}]\,4f^{14}\,5d^{9}\,6s^{1}} \]
GPT-5.6 Luna with reasoning on (Think)Correct
The ground-state electron configuration of a neutral platinum atom (Pt, \(Z=78\)) is: \[ \boxed{[\mathrm{Xe}]\,4f^{14}\,5d^{9}\,6s^{1}} \]
GPT-5.5 (flagship) with reasoning onCorrect
The ground-state electron configuration of neutral platinum (Pt) is: **[Xe] 4f¹⁴ 5d⁹ 6s¹**
14. Gadolinium · Z = 64
“What is the ground-state electron configuration of a neutral gadolinium atom? Give it in noble-gas shorthand.”
elementapi
[Xe] 4f7 5d1 6s2
Source: PubChem Periodic Table, National Center for Biotechnology Information (NCBI), U.S. National Library of Medicine. Record: Gd.json rev. 2026-10-13.
gpt-5.6-luna
Correct
The ground-state electron configuration of neutral gadolinium (Gd, atomic number 64) is: \[ \boxed{[\mathrm{Xe}]\,4f^{7}5d^{1}6s^{2}} \]
GPT-5.6 Luna with reasoning on (Think)Correct
The ground-state electron configuration of neutral gadolinium (Gd, \(Z=64\)) is: \[ \boxed{[\mathrm{Xe}]\,4f^{7}\,5d^{1}\,6s^{2}} \]
GPT-5.5 (flagship) with reasoning onCorrect
The ground-state electron configuration of neutral gadolinium (Gd) is: \[ \boxed{[\mathrm{Xe}]\,4f^7\,5d^1\,6s^2} \] Gadolinium has atomic number 64.
15. Uranium · Z = 92
“What is the ground-state electron configuration of a neutral uranium atom? Give it in noble-gas shorthand.”
elementapi
[Rn] 5f3 6d1 7s2
Source: PubChem Periodic Table, National Center for Biotechnology Information (NCBI), U.S. National Library of Medicine. Record: U.json rev. 2026-10-13.
gpt-5.6-luna
Correct
The ground-state electron configuration of neutral uranium (U, atomic number 92) is: \[ \boxed{[\mathrm{Rn}]\,5f^3\,6d^1\,7s^2} \]
GPT-5.6 Luna with reasoning on (Think)Correct
The ground-state electron configuration of neutral uranium (U, \(Z=92\)) is: \[ \boxed{[\mathrm{Rn}]\,5f^{3}\,6d^{1}\,7s^{2}} \]
GPT-5.5 (flagship) with reasoning onCorrect
The ground-state electron configuration of neutral uranium (U, Z = 92) is: \[ \boxed{[\mathrm{Rn}]\,5f^3\,6d^1\,7s^2} \]

Half-life

How we judged: Correct if the value is within the cited uncertainty or within 3% of the cited value, whichever is wider. Any unit is accepted. Right about stability but outside that tolerance is partly correct; calling an unstable nuclide stable, or vice versa, is incorrect.

16. Bismuth · Z = 83
“Is bismuth-209 stable? If not, what is its half-life?”
elementapi
²⁰⁹Bi is not stable. Half-life 2.01 × 10¹⁹ years (± 0.08 × 10¹⁹)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Bi.json rev. 2026-10-13.
gpt-5.6-luna
Partly correct
Bismuth-209 is **not truly stable**, but it is extraordinarily long-lived. It undergoes alpha decay to thallium-205 with a half-life of approximately: **\(1.9 \times 10^{19}\) years** That is about a **billion times the age of the universe** (~\(1.38 \times 10^{10}\) years), so it is effectively stable for most practical purposes.
Why: Right that it is not stable. 1.9 × 10^19 years is the 2003 measurement; the current evaluation is 2.01 ± 0.08 × 10^19 years, and 1.9 falls just outside it (5.5% low).
GPT-5.6 Luna with reasoning on (Think)Correct
Bismuth-209 is **not truly stable**: it undergoes extremely rare alpha decay to thallium-205. Its half-life is approximately **\(2 \times 10^{19}\) years**—about a billion times the age of the Universe—so it behaves as stable for nearly all practical purposes.
Why: Given to one significant figure, but 2 × 10^19 years lies inside the evaluation's uncertainty (2.01 ± 0.08).
GPT-5.5 (flagship) with reasoning onPartly correct
No. **Bismuth-209 is not truly stable**; it is **radioactive**, undergoing **alpha decay** to thallium-205. Its half-life is about: **\(1.9 \times 10^{19}\) years** That is roughly **19 billion billion years**, over a billion times the current age of the universe, so for most practical purposes it behaves as if it were stable.
Why: Right that it is not stable. 1.9 × 10^19 years is the 2003 measurement; the current evaluation is 2.01 ± 0.08 × 10^19 years, and 1.9 falls just outside it (5.5% low).
17. Tellurium · Z = 52
“Is tellurium-128 stable? If not, what is its half-life?”
elementapi
¹²⁸Te is not stable. Half-life 2.25 × 10²⁴ years (± 0.09 × 10²⁴)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Te.json rev. 2026-10-13.
gpt-5.6-luna
Correct
Tellurium-128 is **not strictly stable**. It undergoes an extremely rare **double-beta decay** to xenon-128: \[ ^{128}\mathrm{Te} \rightarrow ^{128}\mathrm{Xe} + 2e^- + 2\bar{\nu}_e \] Its measured half-life is approximately: \[ \boxed{T_{1/2} \approx 2.2 \times 10^{24}\ \text{years}} \] Because this is vastly longer than the age of the Universe (~\(1.38\times10^{10}\) years), it is effectively stable in ordinary contexts.
GPT-5.6 Luna with reasoning on (Think)Correct
Tellurium-128 is **not truly stable**. It undergoes double beta-minus decay to xenon-128: \[ ^{128}\mathrm{Te}\rightarrow {}^{128}\mathrm{Xe}+2e^-+2\bar{\nu}_e \] Its half-life is approximately: \[ \boxed{2.2\times 10^{24}\ \text{years}} \] This is vastly longer than the age of the universe, so it is effectively stable for ordinary purposes.
GPT-5.5 (flagship) with reasoning onCorrect
Tellurium-128 is **not truly stable**. It is a primordial radioactive isotope that decays by **double beta decay** to xenon-128. Its half-life is extremely long: **≈ 2.2 × 10²⁴ years** That is about **160 trillion times the age of the universe**, so for most practical purposes it behaves as stable.
18. Rubidium · Z = 37
“What is the half-life of rubidium-87? Give one number with units.”
elementapi
⁸⁷Rb is not stable. Half-life 4.97 × 10¹⁰ years (± 0.03 × 10¹⁰)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Rb.json rev. 2026-10-13.
gpt-5.6-luna
Correct
\(4.88 \times 10^{10}\) years
Why: Older tabulated value. The current evaluation is 4.97 ± 0.03 × 10^10 years; 4.88 is 1.8% low, inside the 3% tolerance but outside the evaluation's own uncertainty.
GPT-5.6 Luna with reasoning on (Think)Correct
\(4.88 \times 10^{10}\ \text{years}\)
Why: Older tabulated value. The current evaluation is 4.97 ± 0.03 × 10^10 years; 4.88 is 1.8% low, inside the 3% tolerance but outside the evaluation's own uncertainty.
GPT-5.5 (flagship) with reasoning onCorrect
4.88 × 10^10 years
Why: Older tabulated value. The current evaluation is 4.97 ± 0.03 × 10^10 years; 4.88 is 1.8% low, inside the 3% tolerance but outside the evaluation's own uncertainty.
19. Samarium · Z = 62
“What is the half-life of samarium-147? Give one number with units.”
elementapi
¹⁴⁷Sm is not stable. Half-life 1.066 × 10¹¹ years (± 0.005 × 10¹¹)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Sm.json rev. 2026-10-13.
gpt-5.6-luna
Correct
\(1.06 \times 10^{11}\) years
Why: Older tabulated value. The current evaluation is 1.066 ± 0.005 × 10^11 years; 1.06 is 0.6% low, inside the 3% tolerance but outside the evaluation's own uncertainty.
GPT-5.6 Luna with reasoning on (Think)Correct
\(1.06 \times 10^{11}\ \text{years}\)
Why: Older tabulated value. The current evaluation is 1.066 ± 0.005 × 10^11 years; 1.06 is 0.6% low, inside the 3% tolerance but outside the evaluation's own uncertainty.
GPT-5.5 (flagship) with reasoning onCorrect
1.06 × 10¹¹ years
Why: Older tabulated value. The current evaluation is 1.066 ± 0.005 × 10^11 years; 1.06 is 0.6% low, inside the 3% tolerance but outside the evaluation's own uncertainty.
20. Vanadium · Z = 23
“What is the half-life of vanadium-50? Give one number with units.”
elementapi
⁵⁰V is not stable. Half-life 2.71 × 10¹⁷ years (± 0.13 × 10¹⁷)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: V.json rev. 2026-10-13.
gpt-5.6-luna
Incorrect
\(2.22 \times 10^{17}\) years
Why: The current evaluation is 2.71 ± 0.13 × 10^17 years; 2.22 × 10^17 is 18% low.
GPT-5.6 Luna with reasoning on (Think)Incorrect
\(1.5 \times 10^{17}\) years
Why: The current evaluation is 2.71 ± 0.13 × 10^17 years; 1.5 × 10^17 is 45% low.
GPT-5.5 (flagship) with reasoning onIncorrect
1.5 × 10^17 years
Why: The current evaluation is 2.71 ± 0.13 × 10^17 years; 1.5 × 10^17 is 45% low.
21. Lutetium · Z = 71
“What is the half-life of lutetium-176? Give one number with units.”
elementapi
¹⁷⁶Lu is not stable. Half-life 3.701 × 10¹⁰ years (± 0.017 × 10¹⁰)
Source: NUBASE 2020 evaluation of nuclear properties (2021), AMDC / Chinese Physics C. Record: Lu.json rev. 2026-10-13.
gpt-5.6-luna
Correct
\(3.78 \times 10^{10}\) years
Why: Older tabulated value. The current evaluation is 3.701 ± 0.017 × 10^10 years; 3.78 is 2.1% high, inside the 3% tolerance but outside the evaluation's own uncertainty.
GPT-5.6 Luna with reasoning on (Think)Correct
\(3.78 \times 10^{10}\) years
Why: Older tabulated value. The current evaluation is 3.701 ± 0.017 × 10^10 years; 3.78 is 2.1% high, inside the 3% tolerance but outside the evaluation's own uncertainty.
GPT-5.5 (flagship) with reasoning onCorrect
3.78 × 10¹⁰ years
Why: Older tabulated value. The current evaluation is 3.701 ± 0.017 × 10^10 years; 3.78 is 2.1% high, inside the 3% tolerance but outside the evaluation's own uncertainty.

Standard atomic weight

How we judged: Correct if at least five significant figures are given and the value agrees with the cited standard atomic weight when both are rounded to the shorter of the two precisions.

22. Cesium · Z = 55
“What is the standard atomic weight of caesium, to as many decimal places as you know?”
elementapi
132.905452 u
Source: PubChem Periodic Table, National Center for Biotechnology Information (NCBI), U.S. National Library of Medicine. Record: Cs.json rev. 2026-10-13.
gpt-5.6-luna
Correct
The standard atomic weight of caesium (Cs) is **132.90545196(6) u**. The digits are exact in the currently defined value for the isotope **caesium-133**, which is the only naturally occurring, stable caesium isotope.
GPT-5.6 Luna with reasoning on (Think)Correct
The standard atomic weight of caesium is: **132.90545196(6) u** That is, **132.90545196 unified atomic mass units**, with an uncertainty of approximately **±0.00000006 u**.
GPT-5.5 (flagship) with reasoning onCorrect
The standard atomic weight of caesium is: **132.90545196(6)** That is, approximately: **132.90545196 u** The number in parentheses indicates the uncertainty in the last digits: ±0.00000006.
23. Gold · Z = 79
“What is the standard atomic weight of gold, to as many decimal places as you know?”
elementapi
196.96657 u (± 0.00004)
Source: IUPAC CIAAW Standard Atomic Weights 2021 (2021), International Union of Pure and Applied Chemistry. Record: Au.json rev. 2026-10-13.
gpt-5.6-luna
Correct
The standard atomic weight of gold (Au) is **196.966 570(4)**, commonly written as **196.96657 u**. The **(4)** indicates an uncertainty of ±0.00004 in the last reported digits.
GPT-5.6 Luna with reasoning on (Think)Correct
The standard atomic weight of gold is: **196.966570(4) u** That is, approximately **196.966570 g/mol**. The number in parentheses indicates an uncertainty of **±0.000004** in the last digits.
GPT-5.5 (flagship) with reasoning onCorrect
The standard atomic weight of gold (Au) is approximately: **196.966569 u** Equivalently, its molar mass is about: **196.966569 g/mol** Some references give it as **196.966570(4)**, meaning an uncertainty of about ±0.000004.
Why: 196.966569 rounds to the cited 196.96657; the reply also quotes the IUPAC value 196.966570(4).

First ionization energy

How we judged: Correct if within 1 kJ/mol (0.01 eV) of the cited value in either unit.

24. Mercury · Z = 80
“What is the first ionization energy of mercury? Give one number in kJ/mol or eV.”
elementapi
1007.1 kJ/mol (10.438 eV)
Source: PubChem Periodic Table, National Center for Biotechnology Information (NCBI), U.S. National Library of Medicine. Record: Hg.json rev. 2026-10-13.
gpt-5.6-luna
Correct
10.44 eV (≈1007 kJ/mol)
GPT-5.6 Luna with reasoning on (Think)Correct
1007 kJ/mol
GPT-5.5 (flagship) with reasoning onCorrect
1007 kJ/mol
25. Lead · Z = 82
“What is the first ionization energy of lead? Give one number in kJ/mol or eV.”
elementapi
715.6 kJ/mol (7.417 eV)
Source: PubChem Periodic Table, National Center for Biotechnology Information (NCBI), U.S. National Library of Medicine. Record: Pb.json rev. 2026-10-13.
gpt-5.6-luna
Correct
715.6 kJ/mol
GPT-5.6 Luna with reasoning on (Think)Correct
715.6 kJ/mol
GPT-5.5 (flagship) with reasoning onCorrect
715.6 kJ/mol

How this was done, and what it does not show

Which model. The main column is GPT-5.6 Luna, which OpenAI made the default model for ChatGPT's Free and Go plans in August 2026 (MacRumors, 6 Aug 2026). We asked it with reasoning turned off, which is what happens when you type a question into the free tier and press enter without pressing Think. The other rows show the same model with reasoning on, and OpenAI's flagship GPT-5.5 with reasoning on, so you can see the best case too.

Through the API, not the app. Answers were collected through the OpenAI API, one request per question, with the prompt as the only message and provider defaults. That is the same model the app serves, but the app may wrap it in its own instructions and tools, so a session there can differ. Answers are shown exactly as returned, including the model's own markup, which the app would render.

No cherry-picking. The twenty-five questions and the judging rules were written before any answer was collected, every answer is shown including the ones the model got right, and each question was asked once. Chatbot answers vary from run to run; rerun the prompts yourself.

Disclosure. The half-life rule was widened after the answers were collected, from "within the cited uncertainty" to "within the cited uncertainty or 3%, whichever is wider", because the strict rule would have marked a value 0.6% from the evaluation as wrong. In the main column that change moved three answers (rubidium-87, samarium-147, lutetium-176, all older tabulated values) from incorrect to correct. Every other rule is as first written.

What we make of it. A good model gets most of these right. The ones it missed were delivered with exactly the same confidence as the ones it got right, several of the "right" answers are values from older tables that a stricter rule would fail, and nothing in any answer tells a teacher which is which. That is the case for a citation on every number, not a case against chatbots.

Why these questions. Isotopic abundances, ground-state configurations, long half-lives, standard atomic weights and ionization energies are the kind of values that end up in a quiz distractor or an answer key, where being off by one isotope or one subshell is the difference between a fair question and a wrong one.

Where our numbers come from. The IUPAC isotopic-composition and atomic-weight evaluations, the NUBASE 2020 nuclear-properties evaluation, and PubChem's periodic table, with the specific source cited on each value in the record. While building this dataset we found three isotopic abundances mis-stated in the NUBASE 2020 text table (tellurium-126, lanthanum-138, lanthanum-139) and corrected them against IUPAC; the corrections are cited per value in the records. We mention it because the point of this page is not that we are infallible. It is that every number here has a source you can check.

Quizzes built on these records.

elementapi generates multiple-choice and free-response items from the same cited dataset, with NGSS and AP Chemistry codes on every item. Sign in with your email and try it on the free tier.

Try it free See pricing