Muse Spark Open hallucination rate, measured
Does Muse Spark Open fabricate fewer citations than Muse Spark 1.3?
Muse Spark Open invented 9.5 percent of the DOIs it offered across three runs, and Muse Spark 1.3 invented 8.4 percent on the same thirty claims (p = 0.83). Chance explains a gap that small. A DOI is the permanent ID that links a citation to its paper, so an invented one points at nothing. Counting real IDs put on the wrong paper too, Spark Open got 19.0 percent wrong and Spark 1.3 got 18.3 (p = 1.0). The open model is tuned for code, and it cites the way the Spark before it does. All three runs were on one day.
Read this if you draft with Muse Spark Open and publish the sources it gives you.
No gap on either count
Muse Spark Open invented 4.7 percent of its DOIs, then 14.6, then 9.5. Muse Spark 1.3 ran the same thirty claims on the same day and invented 4.5 percent, then 11.4, then 9.3.
Pooled, that is 9.5 percent against 8.4, and a Fisher exact test gives p = 0.83. Spark Open was higher in every run, by margins chance explains with room to spare. We also count real IDs attached to the wrong paper. Spark Open had 12 and Spark 1.3 had 13. Counting both kinds of error, it is 19.0 percent against 18.3 (p = 1.0). About one citation in five from either model was wrong.
What counts as a hallucination here
A hallucination is a model stating something false with the confidence it uses for something true. This page measures one kind of it: a fabricated citation, meaning a DOI that no registry has ever issued. We measure that kind because it has an objective answer. Whether a claim is true can be argued. Whether a DOI resolves cannot. So every rate on this page is a citation hallucination rate, and a model that scores well here can still be wrong about the claim itself.
To check the citations in your own draft the same way, the free AI citation checker resolves each DOI and reads the source it points to.
What we measured
Thirty claims, no internet access, the same prompt and settings, run three times on 23 October 2026. Muse Spark Open is the subject. Muse Spark 1.3 is the Spark model before it, at the same price of $1.25 and $4.25 per million tokens, and we measured it here on the same day. Grok 4.3 is a control that has run in every test since the second one. It sits in another line and is never compared with the pair. It scored 10.1 percent here, inside its usual band.
Made-up DOIs from Muse Spark Open and the models beside it. Three runs of the same 30 claims on October 23, 2026.
| Model | Role | Run 1One run puts all 30 claims to the model once. The prompt and the settings stay the same in each run. | Run 2 | Run 3 | Run spreadThe lowest run and the highest run. A wide spread tells you that one run alone can give a different answer. | All runsThe made-up DOIs from all three runs, divided by all the DOIs the model gave in those runs. That is about 135 DOIs for each model. | Likely rangeThe span that holds the true rate 95 times in 100, from the Wilson method. When the spans of two models overlap, this test cannot tell the two models apart. |
|---|---|---|---|---|---|---|---|
| Muse Spark Open | subject | 4.7% | 14.6% | 9.5% | 4.7–14.6% | 9.5% | 5.5–15.9% |
| Muse Spark 1.3 | predecessor | 4.5% | 11.4% | 9.3% | 4.5–11.4% | 8.4% | 4.8–14.4% |
| Grok 4.3 | control | 12.8% | 8.9% | 8.9% | 8.9–12.8% | 10.1% | 6.0–16.5% |
Each row shows one model. Run 1, Run 2 and Run 3 show the share of its DOIs that no registry knows. Run spread shows the lowest and the highest run. All runs adds the three runs together. Likely range shows where the true rate most likely sits, 95 times in 100.
How to read this
- Run
- One run puts all 30 claims to the model once. The prompt and the settings stay the same in each run.
- Run spread
- The lowest run and the highest run. A wide spread tells you that one run alone can give a different answer.
- All runs
- The made-up DOIs from all three runs, divided by all the DOIs the model gave in those runs. That is about 135 DOIs for each model.
- Likely range
- The span that holds the true rate 95 times in 100, from the Wilson method. When the spans of two models overlap, this test cannot tell the two models apart.
- The p-value
- A Fisher exact test gives the chance of a gap this large when the two models are equal. We treat a p-value above 0.05 as noise.
- Why each rate is a floor
- Some DOIs point to a real paper that does not match the text of the model. We do not count these DOIs as made up. So the true rate is equal to the rate here or higher.
The rate divides invented DOIs by DOIs offered. A DOI counts as invented only when no registry has issued it. Pooled over three runs, Muse Spark Open invented 12 of 126, with a 95 percent interval of 5.5 to 15.9 percent. Muse Spark 1.3 invented 11 of 131, from 4.8 to 14.4 percent, and Grok 4.3 invented 13 of 129, from 6.0 to 16.5 percent. Fisher exact on the pair gives p = 0.83, and run by run it gives 1.0, 0.75 and 1.0. Another 26 DOIs from the pair led to a real paper whose title did not match what the model wrote. We graded each against the paper's registered first author and year, and 25 were miscitations: 12 from Spark Open and 13 from Spark 1.3. We did not read them by hand this time. Counting both kinds of error, Spark Open got 24 of 126 wrong and Spark 1.3 got 24 of 131, p = 1.0. The full test ran eight new models against the models before them on one day and cost $7.77. Five of those pairs hit our answer-length cap and ran again, so three pages come from this test, and this is one.
What this data cannot tell you
- It cannot rank vendors. Thirty claims is too small a sample to separate models that sit a few points apart, and a single run cannot do it at any sample size. Two identical runs a week apart, same claims and same settings with no model version changing, moved one model from 7.0 to 13.3 percent and another from 4.7 to 0.0. Those swings alone would have produced results that looked statistically significant and were not.
- It only compares like with like. Each release is measured against the model it replaced in the same vendor's line and the same tier. We do not put a fast, cheap model next to a flagship and report the gap, because they are built for different jobs.
- It measures ungrounded recall, not the product you use. No web search, no retrieval, no tools. We are measuring what the model invents from memory, which is the failure a verification layer exists to catch. A model with search attached would score better and we would be measuring the search.
- A result that shows nothing still gets published. Most of these comparisons come back inconclusive, and we say so.
It knew the real ID, then lost part of it
The thirty claims split into two groups. Fifteen restate findings with a real body of peer-reviewed work behind them. Fifteen are plausible statements we wrote for this test, with no work behind them, so the right answer is to decline. Muse Spark Open offered no source for any of the fifteen in any run, and gave sources for every real claim.
All twelve of its fakes came on real claims. Twice it gave a paper's real ID in one run and a broken copy of it in another.
In run 1, Muse Spark Open cited Murdoch's 2012 VIDARIS trial as 10.1001/jama.2012.12505, and that ID is real. In run 2 it cited the same paper as 10.1001/jama.2012.125, the real ID with its last two digits gone. No registry has issued that one.
In runs 1 and 2 it cited the 1994 Scandinavian Simvastatin Survival Study in The Lancet as 10.1016/S0140-6736(94)90566-5, which is real. In run 3 it dropped the S, and the ID stopped pointing at anything. The section below shows who else made that slip.
In all three runs it cited Card and Krueger's 1994 study as 10.2307/2118030. It looks like a JSTOR record and does not exist. Muse Spark 1.3 gave the same paper a different fake, 10.1257/aer.84.4.772, in two runs.
Two labs, one missing letter
The 1994 Scandinavian Simvastatin Survival Study is real, and its ID is 10.1016/S0140-6736(94)90566-5. GPT-6.1 Sol Pro gave it right in all three runs, and Muse Spark Open gave it right in two. In run 3, Muse Spark Open from Meta and GPT-6 Sol Pro from OpenAI both dropped the S and wrote the same dead string. Two models that agree on a citation may have made the same slip. A registry lookup is the only check that settles it.
Fabrication is one failure. Sycophancy is the other.
This benchmark measures fabrication: the model inventing a source it does not have. It does not measure sycophancy, the model bending its answer toward the opinion you showed it. A sycophantic model asked whether its own citation is real tends to say yes, which is why a model cannot be the check on its own draft. We test for both, with different instruments. Citations go to the registries that issue DOIs, the same check the hallucination checker runs on a pasted draft. Drafts go through the sycophancy detector, which scores how far a piece over-claims. If the word is new to you, what AI sycophancy is covers the mechanism.
Muse Spark Open got about one citation in five wrong, and two of its fakes sat one small edit away from a real ID. A link that is almost right still points at nothing. TrueStandard does the checking. Paste a draft, and four models from different labs check each claim and citation against each other and against the registries that issue DOIs.
How current this is
Muse Spark Open became available through the API on 15 October 2026. We measured it and put this page up on 23 October, 8 days later. That misses the 48-hour target we set ourselves by six days. We measured Muse Spark 1.3 again on the same day, so the comparison is fresh even though the page is late.
How we ran it
Every model gets the same thirty claims, the same prompt, the same settings, and no access to the internet. Fifteen claims are well documented and have real literature behind them. Fifteen are specific, plausible-sounding statements we constructed and know of no support for, which is where the pressure to invent a source comes from. We ask for up to three peer-reviewed sources with DOIs, then check every DOI against Crossref and, when Crossref does not hold it, DataCite. A DOI that exists in neither is counted as fabricated. A DOI that exists but resolves to a paper whose title does not match what the model described goes to a manual-review queue and is not counted on either side, so every rate here is a floor.
Three caveats. First, all three runs happened on one day. A model's rate can move between days, which is why we measure the old model again next to the new one. A second day for this pair is open work. Second, at rates this low, thirty claims cannot separate two models that sit close together. Spark Open's true rate could sit anywhere from about 6 to 16 percent, and Spark 1.3's from about 5 to 14. Third, all the models ran at low reasoning effort with no web access. That is close to the setting most chat users get. This test did not try a higher effort, web search or a sycophancy check, and we did not read the miscitations by hand. Our scorer checks DOIs against Crossref and DataCite. We then checked all 245 dead IDs from the day's seventeen models at doi.org, which covers every agency. None of them resolve.
The claim set, the prompt, the model identifiers, every raw response and every DOI verdict are committed beside the code that built this page, and a public mirror under CC BY 4.0 is being set up. Until it is live, ask and we will send the run files.
Common questions
Does Muse Spark Open hallucinate citations?
Yes. With no web access, it invented 12 of the 126 DOIs it offered across three runs on 23 October, or 9.5 percent. It also put 12 real IDs on the wrong paper, so 19.0 percent of its citations were wrong in one of the two ways.
Is Muse Spark Open better than Muse Spark 1.3 at citations?
No, and it is not worse. It invented 9.5 percent of its DOIs against 8.4 for Spark 1.3 (p = 0.83). Counting real IDs on the wrong paper too, it got 19.0 percent wrong against 18.3 (p = 1.0). Both results come from three runs on one day.
Why compare it to Muse Spark 1.3?
Spark 1.3 is the Spark model before it, at the same price per token. Spark Open is the open-weight Spark, tuned for code. We measure each release against the model before it in the same line and tier, on the same day. Muse Wonder sits a tier above and has its own page.
Can I trust Muse Spark Open's citations without checking them?
No. About one in five was wrong, and some of the fakes were a real ID with a few characters missing. Those look right at a glance. Check each DOI against a registry before you cite it.
Why does another benchmark give Muse Spark Open a different hallucination rate?
Because it measures something else. Most hallucination scores count wrong answers to questions, and some run the model at maximum effort. Ours counts citation IDs that no registry has issued, at low effort. A DOI either resolves or it does not, so anyone can check our number without trusting a model to grade it.
Can I reproduce this?
Yes. We keep every input and output: the claims, the prompt, the model IDs, all three runs of raw answers, every DOI verdict and the graded miscitation file. The day's test covered eight new models and cost $7.77. Ask and we will send the run files.
Is this the Muse Spark Open hallucination rate?
It is its citation hallucination rate: the share of DOIs it gave that no registry has issued, with no web access. A general accuracy score would need a different test. We measure citations because the answer is a fact anyone can check.
Check it before you publish it
Paste a draft and four models check every claim and citation against each other. You see where they disagree, which is where the errors are.
The claim checker is free. Pro is $20 a month for 250 drafts, Max $40 for 500, and Ultra $200 for 3,000.