Next-Generation Probiotics: What You Can Actually Buy

You read a study, or a microbiome test result came back with a bar chart, and one organism stood out. Bacteroides vulgatus. Prevotella. Oscillibacter. Blautia. So you searched for a supplement — and found a wall of pages that never quite answer the question.

Here is the answer, stated plainly. Of the roughly nineteen gut organisms people most often try to buy after reading about them, four are purchasable in any honest sense. The rest are not. And for at least three of them, the more useful message is that you would not want a species-level supplement even if one existed.

This page is a September 2026 snapshot of what exists, what does not, and why. Every product detail below is attributed to a specific source we retrieved. Where a manufacturer does not publish a dose, a strain designation or a price, we say so rather than filling the gap.

Three things to understand before you shop

1. AFU is not CFU

Most next-generation products are labelled in AFU (active fluorescent units) or TFU (total fluorescent units) rather than CFU. Pendulum's own consumer explainer describes AFU as a method that counts all live, active cells as well as dead and "in-between" cells.

That distinction matters. A conventional CFU figure counts cells capable of forming a colony — a viability measure. A count that deliberately includes non-viable cells is not interchangeable with it. The category moved to flow-cytometry counting largely because these strict anaerobes do not form colonies well on plates in the first place. If you are used to comparing products by the number on the front of the box, read our CFU guide and then note that here you are often not comparing like with like.

2. "Pasteurized" means not alive

The Akkermansia ingredient authorised in the European Union is specified as pasteurised cells at fewer than ten viable cells per gram, according to the UK Food Standards Agency's published assessment of the ingredient (application RP1468). That is a postbiotic, not a probiotic — even though it sits on the shelf beside probiotics and, as it happens, holds the strongest regulatory file of anything in this category^[1].

3. Label accuracy in this category is unverified

Independent testing of conventional probiotics has repeatedly found gaps between label and contents. A survey of 52 commercial probiotic products documented composition biases, including products falling below their labelled cell counts and products with incorrect species labelling^[2]. A later survey published in the Journal of Functional Foods reported label and content mismatches in 17 of 44 products.

We found no equivalent independent verification of next-generation strain products. That is not an accusation against any brand; it is a gap in the evidence, and it is why we treat unverifiable label claims below as unverified rather than repeating them as fact.

Why the bacteria you read about are mostly not on shelves

Four obstacles stack up, and most organisms hit more than one.

They die in air. This is not a figure of speech. Work published in Nature reported that no colonies of Faecalibacterium prausnitzii DSM 32186 were recovered after 20 minutes of air exposure^[3]. Building a product around an organism like that means anaerobic fermentation, anaerobic drying, anaerobic filling and oxygen-excluding packaging — a different factory from the one that makes a Lactobacillus capsule. A 2025 review of Christensenella minuta names oxygen sensitivity during production, alongside strain-specific effects, as the core obstacles for that organism^[4].

They have no regulatory status. In the US, a novel ingredient generally needs either a New Dietary Ingredient notification or a GRAS determination. In the EU, an organism not consumed to a significant degree before 1997 needs Novel Food authorisation. Most organisms on this page have neither. Note too that "self-affirmed GRAS" — a phrase you will see on several products in this category — describes a determination the company commissioned itself. It is not FDA approval.

The drug pathway often fits better. A defined live organism intended to treat, cure or prevent a disease is a Live Biotherapeutic Product, regulated as a drug. That means an IND, clinical trials and years of work, which is why several of the most promising organisms sit in a pharmaceutical pipeline rather than a supplement bottle.

Some of them, you would not want. This is the part that supplement content rarely says out loud. Several of the most-searched genera contain organisms on opposite sides of the health ledger, and a genus-level or species-level supplement would be a poor idea even if the manufacturing problem were solved.

What about the probiotics you can already buy everywhere?

The familiar genera — Lactobacillus, Bifidobacterium, Bacillus spore-formers, Saccharomyces boulardii — are a different manufacturing and regulatory category from the strict anaerobes discussed here, and this page is not about them. If that is what you were looking for, start with our guide to the best probiotic strains or the overview of probiotic bacteria.

What is actually available right now

Akkermansia muciniphila: available in two forms that are not equivalent

Akkermansia muciniphila is the only organism here with both a real regulatory clearance and a meaningful body of human trial data. It is sold in two distinct forms that people routinely confuse.

Live, US. Pendulum sells A. muciniphila WB-STR-0001. Its single-strain Akkermansia product is listed at 100 million AFU per capsule with chicory inulin — a figure the manufacturer states on its own product page and that independent retailer supplement facts panels match. Pendulum Glucose Control is a five-strain formulation — Clostridium beijerinckii WB-STR-0005, Clostridium butyricum WB-STR-0006, A. muciniphila WB-STR-0001, Eubacterium hallii WB-STR-0008 (now Anaerobutyricum hallii) and Bifidobacterium infantis 100 — with chicory inulin. The company markets it as a medical food for the dietary management of type 2 diabetes under physician supervision, not as an ordinary supplement, and describes its strains as third-party self-affirmed GRAS. A 2022 review in Clinical Diabetes recorded a price of $215 for a 30-day supply^[5] — a four-year-old figure that should be re-checked. Per-strain AFU is not published on the pages we retrieved; a third-party review lists blend totals only, so the dose of any individual strain in these formulas is effectively undisclosed.

Pasteurized, EU and US. The Akkermansia Company (formerly A-Mansia Biotech) sells pasteurised, lyophilised A. muciniphila Muc^T. Its "Akkermansia Essential" consumer page lists 30 billion TFU per capsule, 30 capsules per pack, at $65.00 for a one-time one-month purchase. Its first US product launched on 27 June 2024 as "Healthy Weight" at $59.99 per 30-day supply, pairing the same pasteurised strain with EGCG green tea extract and chromium; it now sells as "Healthy Weight with Glucose Control" at $65.00 one-time on the company's own store, so treat the launch price as historical rather than current.

The regulatory position here is the strongest in the category: pasteurised A. muciniphila was authorised as a Novel Food in the EU on 1 March 2022, capped at 3.4 × 10¹⁰ cells per day for adults^[1], following an EFSA safety opinion^[6]. An amending regulation adopted on 23 February 2026 extended use to the general population from age 12, on a tiered scale — 2.1 × 10¹⁰ cells per day from 12, 3.0 × 10¹⁰ from 14, and 3.4 × 10¹⁰ for adults. Pregnant and lactating women remain excluded: EFSA concluded that safety in that group could not be established, and EU labels must state that the supplement is not for them, nor for children below the relevant age threshold^[7].

On the US side, the FDA's New Dietary Ingredient notification list shows NDI #1438, "Pasteurized Akkermansia muciniphila HB05P," from HealthBiome Inc., submitted 22 September 2025 with an FDA response dated 10 December 2025. The company describes this as an FDA acknowledgement covering a 3.0 × 10¹⁰ cells/day formulation — its second, after NDI #1363 for a lower daily dose in December 2024 — and independent NDI trackers record the response as an acknowledgement rather than an objection. Be clear what that means: an acknowledged NDI notification is the FDA declining to object to a filing. It is not approval of the ingredient, and it is not endorsement of any health claim. HB05P is also a different strain from the ones Pendulum and The Akkermansia Company sell, so its filing says nothing about their products.

And the evidence? This is where the marketing outruns the data, even for the best-supported organism here.

• The 2019 proof-of-concept study everyone cites was explicitly exploratory, with 32 overweight and obese volunteers^[8].
• The multicentre double-blind RCT reporting a 0.6-point A1C reduction and roughly a third lower post-meal glucose AUC versus placebo at 12 weeks tested the five-strain WBF-011 formulation, not Akkermansia alone; its results cannot be attributed to any single organism. It was also designed, funded and authored by the manufacturer, and its primary endpoint — total glucose AUC — reached significance only at p = 0.05^[9].
• A 2026 trial in Nature Medicine (n=90) found that after an 8-week low-energy diet, the pasteurized Muc^T group regained 1.2 kg versus 3.2 kg on placebo over 24 weeks^[10]. Independent experts convened by the Science Media Centre were measured about it, noting the short duration and the need for mechanistic work; one described the result as a clinical and metabolic proof of concept rather than definitive evidence, and another noted that the demonstrated effect on body weight is limited.
• A 2026 trial in Gut Microbes (n=142, metabolic syndrome) missed its primary endpoint: whole-body insulin sensitivity did not differ from placebo after four months of 30 billion pasteurized cells daily in the intention-to-treat analysis. Benefits appeared only in subgroups — participants with low baseline Akkermansia, and postmenopausal women — which is hypothesis-generating, not confirmatory^[11].
• A separate 12-week RCT (n=58) tested a live strain rather than the pasteurized ingredient, and was null overall — body weight and A1C fell similarly in both arms. Benefit appeared only in the pre-specified subgroup starting with low Akkermansia levels, which is what that paper means when it says efficacy depends on baseline levels^[12].

So: buyable, genuinely cleared in the EU, and still one positive weight-maintenance trial, one failed primary endpoint, and a good deal of subgroup analysis. If your interest is metabolic, our metabolic health guide and the metabolic health protocol put that in context.

Anaerostipes caccae: the newest genuine arrival

Anaerostipes caccae CLB101 comes from ClostraBio, Inc., a University of Chicago spinout. The company announced self-affirmed GRAS status on 29 July 2025 and also describes an NDI-exempt designation; a comprehensive toxicological safety evaluation of A. caccae was published in the International Journal of Toxicology in August 2025^[13]. As above, self-affirmed GRAS is a company-commissioned determination, not FDA approval.

Designs for Health launched "Anaerostipes Probiotic," the first commercial product containing CLB101, on 24 September 2025; the strain is also sold direct, and a distribution partnership with Maypro was announced on 12 November 2025. Unlike most of this category, the dose is disclosed: the manufacturer's supplement facts panel lists 30 mg (1 billion AFU) of Anaerostipes caccae CLB101 per capsule, one capsule daily, 30 capsules per container, with delayed-release technology and an instruction to refrigerate on receipt. The manufacturer's page does not display a price; retailer pricing varies. As with every AFU figure on this page, that is a flow-cytometry count, not a CFU count.

The clinical story is unusually transparent for this category, but it is not yet published — and the registry record and the press releases do not agree. NCT07336615 was a randomized, double-blind, placebo-controlled study with a four-week supplementation period; the registry lists 85 participants as actual enrollment, not the 48-subject study named in the company's October 2025 announcement. Nor were the participants simply "healthy": entry required a BMI above 27 and below 35 and self-reported moderate-to-severe gastrointestinal symptoms, scoring 4–7 on the Gastrointestinal Symptom Rating Scale. That is a symptom-selected sample, which matters when reading a gastrointestinal endpoint. Registered secondary endpoints include gastrointestinal symptom scores, serum LPS-binding protein, faecal calprotectin, zonulin, lactoferrin and butyrate, liver enzymes, microbiome sequencing and intestinal permeability. Topline results announced on 22 July 2026 reported that the study met its primary safety and tolerability endpoint and found statistically significant differences in key secondary efficacy endpoints — with the company's own statement noting that full results await peer-reviewed publication. Until that publication exists, the efficacy claim is a company press release, not evidence, and no effect size, endpoint or p-value has been made public. The underlying science is a mouse study in which an A. caccae plus lactulose synbiotic prevented and treated food allergy^[14].

One nomenclature note, because readers conflate them constantly: Anaerostipes caccae and Anaerobutyricum hallii are different organisms. Both produce butyrate, and both were historically shuffled among the Eubacterium and Clostridium clusters.

Faecalibacterium prausnitzii: available since 2026, as a minor component

F. prausnitzii became purchasable in April 2026 through a practitioner-channel product, Designs for Health "Complete Commensal Probiotic," launched 15 April 2026. Retailer panels list five organisms per capsule:

That is a 280 mg blend totalling 720 million AFU per serving, one capsule daily, 30 capsules per container, listed at $79.98 at one retailer and marketed as "single super donor-sourced" with "patented oxygen-free technology."

Three things a reference site has to point out. First, the F. prausnitzii content is 10 million AFU — about 1.4% of the total count; by cell number the product is overwhelmingly Akkermansia and bifidobacteria. Second, AFU includes non-viable cells. Third, we identified no human clinical trial of this product; the launch release cites emerging research generically and carries the standard FDA disclaimer. The identity of the strain supplier or donor behind the TA-series designations is not disclosed on any page we retrieved.

The reason this took so long is the oxygen problem. The Nature group that measured the 20-minute die-off solved it through stepwise oxygen-tolerance adaptation plus co-culture with Desulfovibrio piger, producing a strain that survived freeze-drying and two weeks at −20 °C with almost no loss of viable count^[3]. A related tolerability study, NCT03728868, administered F. prausnitzii with D. piger as a dietary supplement over eight weeks.

Be clear about what the evidence supports. The case for F. prausnitzii rests on depletion association — notably in Crohn's disease — plus preclinical anti-inflammatory work. There are no efficacy RCTs showing that supplementing it improves a human clinical outcome. Availability arrived ahead of evidence.

Anaerobutyricum hallii: available, but only inside a blend

A. hallii, labelled historically as Eubacterium hallii WB-STR-0008, is one of five strains in Pendulum Glucose Control and Metabolic Daily. It is not sold standalone, and per-strain AFU is not published.

The awkward part is that the better human evidence belongs to a sibling species you cannot buy: Anaerobutyricum soehngenii L2-7, developed by Caelus Health. Its four-week dose-finding pilot in men with metabolic syndrome was safe and well tolerated but stated plainly that there was no overall effect on hepatic insulin sensitivity or glucose disposal in either dose group; the positive signal was a correlation between achieved faecal abundance and improvement in peripheral insulin sensitivity, with response depending on baseline microbiota^[15]. Later work reported improved glycaemic control and other cardiometabolic markers in adults at risk of type 2 diabetes^[16]. The hypothesis originated in db/db mice^[17].

So the A. hallii you can buy sits inside a formula whose trial tested the whole formula, and the A. soehngenii with dedicated human trials is not a consumer product.

Organism-by-organism availability

Notes on the ones you cannot buy

Bacteroides vulgatus returns nothing but research literature, test panels that measure it, culture-collection entries and contract-research pages. A social-media post referencing a B. vulgatus product exists but is not a verifiable source, and we are not treating it as evidence that a product exists. The direction of effect is also genuinely contested: the same review literature that discusses probiotic potential in Bacteroides flags that safety work is incomplete^[18].

Bacteroides fragilis splits into non-toxigenic and enterotoxigenic strains, the latter producing fragilysin, a zinc metalloproteinase implicated in barrier disruption. It is also the anaerobe most frequently isolated from human clinical infections. That is a species-level safety fork, which is precisely why a "B. fragilis supplement" would be a poor idea. In China, a fermentation product made from strain BF839 has been approved as a safe food raw material and is described in review literature as extensively used there^[19]; a randomized trial of BF839 in pediatric autism spectrum disorder has been published^[20], though effects on core symptoms across probiotics generally remain unclear by that paper's own framing. We found no US or EU product, NDI or Novel Food authorisation.

Prevotella has no supplement. The lead asset, EDP1815 — an oral, gut-restricted, non-live preparation of P. histicola — completed a Phase 2 trial in plaque psoriasis^[21], but Evelo Biosciences ceased development in atopic dermatitis in 2023 and its board resolved to dissolve the company that November. P. copri is the most contested organism on this list: it was strongly associated with new-onset untreated rheumatoid arthritis and exacerbated colitis in mice in one landmark study^[22], while other cohorts found no association — most plausibly because extreme strain-level heterogeneity means "P. copri" is not one thing^[23]. "More Prevotella" is not a coherent health goal.

Christensenella minuta is a drug candidate, not a supplement. Xla1 (strain DSM 33407) was developed by YSOPIA Bioscience explicitly as a Live Biotherapeutic Product, with FDA IND clearance and a Phase 1 trial, NCT04663139, in healthy-weight, overweight and class-1 obese adults; the preclinical rationale is published^[24]. Consumer listings naming Christensenella do exist on marketplaces — but for the one we examined, the supplement facts panel, strain designations and price did not render, so whether it contains viable C. minuta of a defined strain is unverified. Given the label-accuracy literature and the manufacturing difficulty, that gap matters. The well-known finding for this organism is heritability and association with leanness in twin cohorts, plus mouse studies; human interventional evidence beyond a Phase 1 safety and tolerability trial is, as far as this research found, absent^[4].

Blautia has no product for B. wexlerae or B. obeum. The closest clinical asset is a different species: Blautix (MRx1234), a Blautia hydrogenotrophica live biotherapeutic from 4D pharma, which completed a Phase 2 trial in IBS and was described by its authors as having potential to become a novel, safe treatment option^[25]. For B. wexlerae specifically the evidence is depletion association in obesity, prediabetes and type 2 diabetes plus mouse and mechanistic work^[26]; we found no human trial of B. wexlerae supplementation. Patents claiming Blautia compositions exist, and patents are not products.

Coprococcus eutactus is purchasable only as an ATCC research culture at $486 — a laboratory reagent, not a supplement. The widely cited human data are observational: Coprococcus was depleted in relation to depression and quality-of-life metrics in a large population cohort^[27]. The interventional evidence is a single mouse colitis study^[28]. No developer program was identified.

Ruminococcus is the clearest case of a genus holding organisms on opposite sides of the ledger. A 2024 next-generation-probiotic review tabulates R. bromii as intended for live biotherapeutic development with no clinical trials listed^[29]; its interest is as the keystone resistant starch degrader, which makes it a dietary question rather than a supplement one. R. gnavus, by contrast, is a pathobiont enriched in Crohn's disease that produces an inflammatory glucorhamnan polysaccharide capable of inducing TNF-α^[30]. No responsible product would contain it.

Oscillibacter and Oscillospira have no product. Oscillospira is the textbook uncultured gut genus, and the cholesterol work that made Oscillibacter famous began from uncultured metagenomic species^[31]. The headline result — Oscillibacter species associated with lower blood and stool cholesterol, with cholesterol-metabolizing capability confirmed in isolates in vitro — comes from Framingham Heart Study profiling^[32]. That is association plus in vitro work, with no human intervention. Press coverage at the time was explicit that it merely hinted at future probiotics.

Turicibacter evidence is entirely animal work: strains differentially modify bile acids and host lipid metabolism via bile salt hydrolases^[33], and T. sanguinis interacts with host serotonin transport. We found no human data of any kind on Turicibacter supplementation. See secondary bile acids for what that mechanism actually involves.

Odoribacter sits at candidate-review stage^[34]. O. laneus improved glucose control in obese mice by depleting circulating succinate^[35]. Published work on a time-dependent colon-delivery system for O. splanchnicus is itself a signal that delivering this strict anaerobe remains an unsolved engineering problem. No human trials found.

Parabacteroides distasonis has no product, no NDI and no Novel Food authorisation, and this is a safety story as much as a manufacturing one: review literature describes the genus as carrying emerging antimicrobial resistance and both pathogenic and probiotic roles, with effects that are strain-dependent in both directions^[36].

Dysosmobacter welbionis is at mouse stage. Strain J115^T prevented diet-induced obesity and metabolic disorders in mice^[37], and a 2025 human paper links faecal Dysosmobacter concentration to the plasma lipidome — observational^[38]. No human intervention trial was identified.

Eubacterium rectale, now reclassified as Agathobacter rectalis, has no product and no program; the evidence is association plus preclinical work. The name change is worth knowing if you searched the old name and found nothing.

What to do instead: feed what you already have

If the organism you want cannot be bought, the remaining lever is diet — with one honest caveat stated up front. We cannot promise that a specific fibre will raise a specific genus in a specific person. Response to fermentable substrate varies between individuals, and several of the trials above found that even a supplemented organism's effect depended on baseline microbiota. Treat the following as reasonable, low-risk, evidence-informed practice rather than a targeting protocol.

• Resistant starch is the substrate most directly tied to an organism on this page: R. bromii is the keystone degrader, which makes resistant starch the most defensible dietary lever for anyone whose test flagged it — though, per the caveat above, feeding a substrate is not the same as raising a genus in a given person.
• Lactulose was the partner substrate in the A. caccae synbiotic work in mice^[14] — useful mechanistic context, but lactulose is a regulated laxative medicine in most markets rather than a food fibre, and a mouse synbiotic study is not a reason to start taking it.
• Chicory inulin is the prebiotic formulated into the Pendulum products described above; inulin and FOS are the most widely used bifidogenic substrates.
• Variety over any single fibre. The broader principle — that a diverse intake of fermentable fibre supports the short-chain fatty acid-producing community, and butyrate producers in particular — is the honest fallback, and it is not organism-specific. Our prebiotic foods guide and the full prebiotic fibre library cover the options, including arabinoxylan, beta-glucan, pectin, GOS and psyllium.

For a structured approach rather than a shopping list, see how to improve your gut microbiome and the gut restoration protocol.

How to read a test result that flags a low organism

Microbiome reports are good at producing an organism name and a red bar. They are much worse at telling you what to do about it. A few principles:

1. Check whether the organism is even buyable before you go looking. For fifteen of the nineteen entries in the table above, there is no verifiable consumer product — so a low reading is information about your ecosystem, not a shopping instruction.
2. A low genus is not a diagnosis. Reference ranges are laboratory-specific, sequencing methods differ, and a single sample captures one moment. Our complete guide to microbiome testing covers what these panels can and cannot tell you.
3. Direction of effect is not always what the report implies. Several organisms here — P. copri, R. gnavus, P. distasonis, and B. fragilis at species level — sit on both sides of the ledger depending on strain. A report that treats a genus as uniformly good or bad is oversimplifying.
4. Baseline matters, in both directions. Two Akkermansia trials found effects that depended on the participant's starting level^[11][12]. That cuts both ways: it is a reason not to assume a product will work, and a reason not to assume it cannot.
5. Bring it to a clinician, especially if the result arrived alongside real symptoms, or if you are considering a product marketed as a medical food — which is by definition intended for use under medical supervision.

What could change this list

The pipeline is real, even if it is thinner than the marketing implies. The oxygen-adaptation work published in Nature showed that a strict anaerobe once considered unmanufacturable can be made to survive freeze-drying and storage^[3] — the technical unlock behind F. prausnitzii reaching a shelf at all. Christensenella is in Phase 1 as a drug. ClostraBio's A. caccae trial has reported topline results and awaits peer-reviewed publication. And on the other side, Evelo's collapse is a reminder that a Phase 2 asset can disappear entirely.

We will update this page as products, authorisations and trial publications appear. Where a fact here is marked unverified, it stays unverified until a primary source says otherwise.

This article is for general educational purposes and is not medical advice. Nothing here should be read as a claim that any supplement treats, cures or prevents any disease. Product details reflect sources retrieved in September 2026 and may change without notice; verify current formulations, doses and prices with the manufacturer. Talk to a qualified healthcare professional before starting any supplement, particularly if you are pregnant, immunocompromised, taking prescription medication, or managing a diagnosed condition.