The three terms are not the same kind of thing
A Broad Agency Announcement is a solicitation method. A Commercial Solutions Opening is a solicitation method. An other transaction is an award instrument. Put plainly: two of them describe how the government asks, and one describes what the parties sign at the end. That is why the comparison feels slippery when people line all three up in a row. A CSO can end in an other transaction. A BAA can end in an other transaction. An other transaction can arrive by way of either, or by way of neither.
Separate the two questions and the decision gets small enough to hold in your head. Question one: what is the government buying, and does the law let it ask this way? Question two: what instrument fits the risk, the intellectual property, and the money? Most bad pathway decisions come from answering only the second question, usually with the word "OTA," and then finding that the first question had a different answer all along.
What follows is the working version of each, current through the statutory changes Congress made in December 2025 and the guidance the department reissued in early 2026.
The Broad Agency Announcement: a general call for research
The BAA is the oldest of the three and the most narrowly scoped. FAR 35.016 sets the procedures. It acquires basic research, applied research, and the part of development not tied to a specific system or hardware procurement, and the regulation describes it as a way to fulfill requirements for scientific study and experimentation directed toward advancing the state of the art or increasing knowledge.
Three mechanics matter to anyone deciding whether to respond. First, a BAA is a competitive procedure under the Competition in Contracting Act, but only when the announcement is broad, open to all capable sources, includes selection criteria, and the award follows a peer or scientific review. That test lives at FAR 6.102(d)(2). Second, because offerors are not responding to a common statement of work, FAR 35.016 says proposals need not be evaluated against each other; selection turns on technical merit, importance to agency programs, and fund availability. Third, agencies publicize BAAs through the governmentwide point of entry no less than annually, which is why office-wide announcements sit open for years with periodic amendments.
Inside the Department of War, the secondary title the department now uses in most non-statutory communications, the BAA carries a restriction the FAR does not impose. DFARS 235.016 confines the peer-reviewed BAA to science and technology proposals in budget activities 6.1 through 6.4: basic research, applied research, advanced technology development, and advanced component development and prototypes. If the money is procurement money or later-stage development money, a BAA is the wrong door regardless of how research-flavored the work sounds.
The BAA is also the most instrument-agnostic of the three. DARPA states plainly that a proposal submitted against one of its announcements can result in a procurement contract, a grant, a cooperative agreement, or an other transaction. The announcement tells you which instruments are on the table for that call. Read that paragraph before you write anything else.
The Commercial Solutions Opening: a general call for something that already works
A CSO is a general solicitation that describes a problem, or an area of interest, and invites companies to propose whatever they think solves it. There is no statement of work and no common baseline against which offers are ranked. Evaluation runs through subject-matter expert review against criteria stated in the announcement, and solutions can be judged on their own merits without trade-offs against each other.
For the Department of War, the authority is 10 U.S.C. § 3458, implemented at DFARS subpart 212.70 and finalized in August 2023. The regulation requires fixed-price contract types, including fixed-price incentive where appropriate, and directs that what is bought this way be treated as commercial products or commercial services. One provision in DFARS 212.7002 is worth memorizing: CSO procedures may be used together with FAR part 35 and DFARS part 235 for research and development, and a CSO used that way is not subject to the 6.1-through-6.4 budget-activity limitation at DFARS 235.016. That sentence is why a program office with development money and a research-shaped problem often reaches for a CSO rather than a BAA.
What December 2025 changed
Public Law 119-60, enacted December 18, 2025, rewrote several subsections of § 3458. The changes are large enough that pre-2026 guidance on CSOs should be read with the amendment date in hand.
The word "innovative" came out. The authority previously reached innovative commercial products and services. As amended, it reaches commercial products, commercial services, and nondevelopmental items. A CSO is no longer a specialty tool reserved for novelty.
The review requirement loosened. Selection may now follow a peer, technical, or operational review, as appropriate. Operational review is a real addition for capability judged by whether it works in the field rather than by whether it is scientifically interesting.
Follow-on production is written into the statute. After a competitive selection under a CSO, the department may enter into a follow-on production contract or transaction, including sole-source. That closes a gap where the production path ran cleanly out of a prototype other transaction under § 4022(f) but less cleanly out of a CSO.
The high-dollar written determination went away. The prior requirement for a written determination above $100 million was removed. Congressional notification survives: the department notifies the defense committees no later than 45 days after awarding a contract exceeding $100,000,000.
The regulation has not caught up. As of this writing, DFARS 212.7001 still defines the work as "innovative," and DFARS 212.7003 still calls for senior procurement executive approval above $100 million. That gap is normal after an authorization act, and it is being worked alongside the broader FAR rewrite, which moved into formal rulemaking in June 2026. When the two texts disagree the statute governs, and the contracting officer's own agency deviation tells you which text that office is actually operating under. Ask.
The civilian CSO is a different animal
Outside the Department of War, CSO authority is a pilot rather than a permanent power. Section 880 of the FY2017 National Defense Authorization Act created a commercial solutions opening pilot program for GSA and DHS. Section 7227 of the FY2023 NDAA amended it, set the total award limitation at $25 million, and carried the pilot to a sunset of September 30, 2027.
The civilian pilot kept the older, tighter definition, limited to innovative commercial products and services, where innovative means any new technology, process, or method, including research and development, or any new application of an existing one. GSA implements the pilot at GSAM part 571, and its guide is explicit that competitive selection from a general solicitation with peer review by government subject-matter experts counts as a competitive procedure. It is also explicit about intellectual property in a way federal guidance rarely is: a stated preference for the company retaining core IP where appropriate, and a warning to contracting officers that burdensome IP terms drive new entrants away.
Two civilian details change how a company should behave. Unsuccessful offerors may request technical feedback within three business days of non-selection, a short clock and a real one, and protests go to the agency under GSAM 533.103 rather than the familiar external route. If the pilot matters to your plan, watch the 2027 sunset. Making the authority permanent for GSA and DHS has been proposed alongside the FAR overhaul; until something passes, the expiration date is the planning assumption.
The other transaction: the instrument, not the ask
An other transaction is an agreement that is not a procurement contract, not a grant, and not a cooperative agreement. Because the FAR applies to procurement contracts, it does not apply here on its own. Anything the FAR would have supplied has to be written into the agreement or it is simply absent.
The department has two OT statutes producing three kinds of award. Research other transactions sit at 10 U.S.C. § 4021 and cover basic, applied, and advanced research. Prototype other transactions sit at § 4022. Production other transactions come out of § 4022(f), where a competitively awarded prototype that was successfully completed can move to production without further competition.
The conditions differ in a way that decides real cases. A prototype OT requires one of four things under § 4022(d): a nontraditional defense contractor or nonprofit research institution participating to a significant extent, or all significant non-federal participants being small businesses or nontraditional contractors, or at least one third of total cost from non-federal sources, or a written determination by the senior procurement executive that exceptional circumstances justify it. Meet the first or second and no cost share is required at all.
A research OT is different, and the difference is money out of your own pocket. The OUSD(R&E) Guide to Research Other Transactions, revised February 12, 2026, states that resource sharing is a key element of the research statute, that to the maximum extent practicable government funds should not exceed the total provided by the other parties, and that the statutory default is generally a 50/50 share. The guide is candid that the final share is flexible and should account for the performer's resources, prior investment, and commercial relevance. Still, the planning assumption is that the company brings something.
Two provisions of § 4021 rarely get mentioned and both help a small company: advance payments are specifically allowed, in any amount and without further approvals, and information submitted during a research OT process can be withheld from public release for five years.
How the three combine
Hold it as a two-by-two: solicitation method on one axis, instrument on the other. A DoW CSO can end in a FAR part 12 commercial contract, a prototype or production other transaction, or a cooperative agreement, depending on what the announcement says and which authority the contracting officer is using. A BAA can end in a contract, a grant, a cooperative agreement, or an other transaction. The Defense Innovation Unit is the best-known pairing: it solicits through commercial solutions openings and principally awards prototype other transactions, running a short solution brief, then a pitch, then negotiation, with the office committing to respond within 30 days on whether it wants the pitch.
The 2026 research OT guide makes the same point from the government's side: organizations use Broad Agency Announcements or Commercial Solutions Openings because those methods accommodate a variety of award instruments, and government teams are free to invent an approach if none of the existing ones fit. The announcement in front of you is the authority on what it will produce.
Which pathway is the right first call, by kind of work
Each number is an editorial confidence that the named pathway is the right first call for that kind of work, read from public statute and agency guidance. Judgement, not measurement.
The bottom row is the honest one. When a requirement is small, well defined, and repeatable, simplified acquisition procedures under the FAR beat all three on elapsed time and nobody negotiates a single term. Pathway flexibility earns its keep where the alternative would have been a full negotiated source selection.
Side by side on the axes that decide
| Axis | Broad Agency Announcement | Commercial Solutions Opening (DoW) | Other transaction |
|---|---|---|---|
| What it is | Solicitation method for research (FAR 35.016) | Solicitation method for commercial items (10 U.S.C. 3458) | Award instrument (10 U.S.C. 4021, 4022, 4022(f)) |
| What it can buy | Research, and development not tied to a specific system; DoW limits it to budget activities 6.1–6.4 | Commercial products, commercial services, nondevelopmental items, since December 2025 | Research, prototypes, and production follow-ons to a completed prototype |
| How offers are judged | Peer or scientific review; proposals need not be compared against each other | Peer, technical, or operational review; solutions may be judged on individual merit | Whatever the solicitation says; guidance asks for fairness, not a fixed process |
| Instruments it produces | Contract, grant, cooperative agreement, or other transaction | FAR part 12 contract, prototype or production OT, cooperative agreement; fixed-price required | Itself; terms negotiated from close to a blank page |
| Cost share | Set by the announcement | Not a feature of the authority | None for a prototype OT with a nontraditional participant; roughly 50/50 default for research |
| Protest posture | Ordinary routes when the award is a procurement contract | GAO has taken CSO protests and reviewed them deferentially | GAO generally declines the award; it will review whether the authority was properly used |
Protest and remedy: three levels of exposure
BAA. When the award is a procurement contract, it is a procurement contract, and the ordinary protest routes are available on ordinary terms. The peer-review structure gives the agency wide discretion on the merits, but the forum is not in doubt.
CSO. GAO has heard them. In PavCon, LLC, B-420640 (July 5, 2022), GAO reviewed an Air Force CSO evaluation, held that the agency was not required to conduct discussions under CSO procedures, and denied the protest, applying a deferential standard suited to research and development selections. The forum exists. The standard of review is not generous.
Other transaction. Agreements awarded under other transaction authority are not procurement contracts, so GAO generally does not review the award or the solicitation. It will review a timely protest arguing that the agency is improperly using other transaction authority to buy goods or services it should be buying under a contract, the holding companies cite from MorphoTrust USA, LLC, B-412711 (May 16, 2016). The Court of Federal Claims is the other door, and the Federal Circuit addressed a challenge tied to a follow-on production award in Oracle America, Inc. v. United States, 975 F.3d 1279 (Fed. Cir. 2020).
Civilian CSO pilot. GSA's guide sends offerors to agency-level protest procedures under GSAM 533.103, with a three-business-day window to request technical feedback after non-selection.
Plan for both directions. Awards with a thin protest surface are durable once you win them and hard to contest when you lose them. That asymmetry should shape how much a company spends chasing any single one.
Flexibility in the acquisition method is not flexibility in the law
Appropriations still have a purpose, a time, and an amount, and the 2026 research OT guide says so directly: other transactions are exempt from most acquisition statutes, but fiscal law applies. Section 889 supply-chain prohibitions and Federal Acquisition Supply Chain Security Act orders reach non-FAR awards as well as FAR ones, which GSA's CSO guide states as an award-eligibility condition. Export control, classification, and controlled unclassified information handling follow the data and the work, not the instrument.
Money, books, and audit
A DoW CSO award must be fixed-price. That is a regulatory requirement, not a preference, and it means the price you propose is the price, with schedule risk sitting on your side of the table. The benefit is that a fixed-price commercial award asks nothing of your accounting beyond ordinary commercial books.
An other transaction goes further. The research OT guide sets the policy plainly: avoid requirements that would force participants onto financial management systems different from what they already use, expecting recognized accounting principles and complete, accurate records rather than a government-approved system. Audits and access to financial records are negotiated, and fixed-support awards generally should not carry an audit provision at all. Payment tends to run on payable milestones tied to demonstrable technical events.
The relief is real and it is temporary. A production follow-on at scale, or a research program that later converts to a cost-reimbursement contract, puts a compliant accounting system back into scope. Companies that skip that build find the gap at the least convenient moment.
Intellectual property
Under a BAA that results in a FAR contract, the standard data rights apparatus attaches and the negotiation is about assertions and markings inside a known framework. Under a CSO that results in a fixed-price commercial contract, the government's rights follow from the commercial license it accepts. Under an other transaction, nothing attaches by default. Every term is written.
The practical advice is the same across all three and it is unglamorous. Bring a background intellectual property schedule to the first negotiation, listing every component, model, library, and dataset that predates the effort. Define the license grant per deliverable rather than across the project. Treat training data, model weights, evaluation sets, and prompt libraries as separate assets, because a clause drafted for "technical data" often does not say which of them it reaches.
What the sequence looks like
A commercial solutions opening, end to end
Sequence drawn from GSA's CSO guide and DIU's published process. Durations vary by office; treat them as shape, not schedule.
One caution about durations, because speed is the most oversold part of this subject. The 2026 research OT guide addresses it head on: agencies often choose an other transaction for the perceived ability to award quickly, and history shows that speed happens only when the government team works together from the earliest planning stages through award. The instrument does not create the schedule. The sponsor, the money, and the agreements officer's queue create the schedule.
Where each one fails
A BAA fails when the work has a customer. If a program office wants a thing delivered on a date, the peer-review structure and the no-common-work-statement premise fight the requirement. Inside the department, the budget-activity limit usually settles it before anyone argues.
A CSO fails when nothing commercial exists yet. Even with the "innovative" limitation gone at the defense level, the authority reaches commercial products, commercial services, and nondevelopmental items. Work that has to be invented from a blank page is a poor fit, and the fixed-price requirement makes an underspecified scope dangerous to price.
An other transaction fails when nobody reads the agreement. The blank page is an advantage only for the party that arrives with drafted terms. The party that arrives without them signs the other side's opening template, and that template is written for the other side.
Bottom line
Answer the questions in this order. Is the work research, or is it a delivery? Research points to a BAA or a research other transaction, and the research OT adds a cost-share expectation the BAA may not. Is there a named program office with money that wants the thing? A yes points to a CSO or a prototype other transaction and makes the production conversation real rather than theoretical. How much does your existing intellectual property matter? The more it matters, the more a negotiated instrument is worth the extra weeks. How much protest protection are you willing to give up? The faster the pathway, the thinner your remedy if you lose.
Four answers, and the pathway usually names itself. The people who struggle with this decision are almost always trying to pick an instrument before deciding what they are buying. And all three fail on the same thing: the pathway controls the terms and the timeline, but it does not make anyone want the capability.
Common objections, answered plainly
Does using a non-FAR pathway mean the rules do not apply?
It means the FAR does not apply. Fiscal law, the Anti-Deficiency Act, Section 889 prohibitions, supply-chain orders, export control, and security requirements all still reach the work. Program offices routinely write FAR-like protections into these agreements by hand, because they want them and the clause that would have supplied them is absent.
Are these pathways actually faster?
The evaluation step usually is, because a short solution brief replaces a full proposal. What does not compress is what controls the calendar: the money has to exist and be the right kind, the requirement needs a sponsor, security paperwork takes what it takes, and the agreements officer has a queue. The department's own research OT guide says speed happens only when the government team is aligned from the planning stage.
Which one should a company new to federal work pursue first?
The one that matches the work, not the one with the best reputation. A company with a working product and a program office that has seen it should be reading commercial solutions openings. A company with a research result and no fielding path should be reading office-wide announcements. A company with neither has a customer discovery problem, and no pathway solves that.
Frequently asked questions
Yes. What a CSO can award depends on what the announcement says and which authority the contracting officer is using. Awards can take the form of FAR part 12 commercial contracts, prototype or production other transactions, or cooperative agreements. The Defense Innovation Unit solicits through commercial solutions openings and principally awards prototype other transactions.
Public Law 119-60, enacted December 18, 2025, removed the "innovative" limitation and extended the authority to commercial products, commercial services, and nondevelopmental items. It allowed peer, technical, or operational review, wrote follow-on production authority into the statute including sole-source awards, and removed the written-determination requirement above $100 million while keeping congressional notification within 45 days of an award over that amount.
Resource sharing is a core element of the research authority. The February 2026 OUSD(R&E) guide states that, to the maximum extent practicable, government funds should not exceed the total provided by the other parties, with a general 50/50 default position and flexibility based on the performer's resources, prior investment, and the commercial relevance of the work. A prototype other transaction requires no cost share when a nontraditional defense contractor participates to a significant extent.
It depends on the instrument. A procurement contract awarded from a BAA or a CSO carries the ordinary routes, and GAO has decided CSO protests under a deferential standard. Agreements awarded under other transaction authority are not procurement contracts, so GAO generally declines to review the award itself, while agreeing to review whether the agency improperly used the authority at all. GSA's civilian CSO pilot directs offerors to agency-level protest procedures.