Avogadro 2026 Survey Results

[ghutchis]

A few weeks ago, I posted the 2026 Community Feedback Survey - and I’d like to start doing this annually.

Thank you to the 49 50 of you who filled out the development survey. Several people wrote multi-paragraph answers!

This was a self-selected sample recruited through this forum, social media, and the project website at https://avogadro.cc/, so it over-represents people who are already engaged. 65% use Avogadro 2 regularly, 78% use Avogadro at least weekly, and only three respondents had never tried v2.

So this is a survey of the active community rather than of everyone who has ever installed Avogadro — the lapsed 1.2 users are almost certainly under-represented. (Of course, these are 49 50 people already active so your comments are important.)

[ghutchis]

Topic	Summary
Region	Europe 45%, North America 16%, Asia/Middle East 14%, Latin America 10%, Africa 6%, Oceania 4%
Role	Grad student 37%, industry/government 18%, postdoc 18%, undergrad 10%, faculty 8%
Platform	Linux 55%, Windows 53%, Mac 27% (multi-select)
Use	Mostly research 76%, both equally 14%, mostly teaching 4%
Frequency	Several times a week 39%, daily 18%, a few times a month 18%, weekly 12%
Install	Downloaded 76%, package manager 41%, compiled 8%

Web search is the single largest way people find Avogadro (29%), ahead of an advisor or instructor (16%) and a friend or peer (14%). However, word-of-mouth together is 38%.

[ghutchis]

Most respondents are long-time users, with 63% using Avogadro for over 5 years:

[ghutchis]

Most respondents are using for research, but only four respondents are faculty, so we’d probably want a separate survey about teaching use (e.g., are there issues with adoption or use in a classroom or teaching lab deployment).

[ghutchis]

Lots of molecular chemistry, including organic, inorganic / organometallic users. Interesting that 11 people mentioned chemical education here, but that doesn’t match the count for teaching / both above.

(Less for polymers, supramolecular, solid-state, etc.)

[ghutchis]

How does Avogadro 2 compare to 1.2?

Respondents rated v2 against 1.2 on a 1–5 scale. Excluding the three who’d never tried v2:

	Mean	Median	Rated 5	Rated ≤2
Speed	3.98	4	16	3
Ease of use	3.68	4	13	8
Stability / bugs	3.70	4	12	6

Speed is the clear win. A third of respondents gave it a 5, and nobody rated it below 2. One long-time user put the rewrite in perspective better than I could — that despite the years it took to reach feature parity, it was the right call.

Ease of use and stability have a tail: the people who rated low wrote the longest free-text answers, and they converge on the same specific complaints (discussed below).

The ratings improve the more you use it

Tried v2	n	Speed	Ease	Stability
Regularly	32	4.03	3.84	3.81
A few times	10	4.00	3.40	3.50
Maybe once or twice	2	3.00	2.50	3.00

To me this suggests some of the negative signal is an outdated impression formed on early builds.

One respondent said so directly:

What I needed Avogadro for, V2 didn’t have that at the time. Don’t know if it still doesn’t.

So hopefully people will continue to update to 2.0 and beyond.

[ghutchis]

About the Apple Silicon comments

Two Mac respondents raised Apple Silicon, and I need to comment – I develop on an M4 MacBook Air, so “no Apple Silicon support” was surprising to say the least. (It’s also pretty clear on Install — Avogadro 2.0.0 documentation

Official Apple Silicon builds exist and are great. If you’re on a Mac and Avogadro is behaving strangely, please try a build from avogadro.cc before assuming it’s the application — and tell us which package you used. (Homebrew should install directly from the website.)

One of these users said specifically they’d only update to 2.0 when Apple stopped supporting Intel binaries. (Soon, evidently.)

[ghutchis]

A lot of you want to help and can’t find the door

63% have never contributed. But only 4 people out of 49 said they wouldn’t help in future.

Would contribute	n	%
Feature suggestions	26	53%
Bug reports	25	51%
Testing pre-release / nightlies	23	47%
Coding	20	41%
Forum participation	16	33%
Documentation / teaching materials	11	22%
Translations	7	14%

Twenty people are willing to write code. The barriers they name are not ability and not motivation:

What would make it easier	%
Knowing what tasks currently need help	53%
A clear list of good first issues	45%
More time (nothing we can change)	41%
Knowing that small contributions are welcome	33%
Better contributor documentation	29%
Academic credit — citation or co-authorship	29%
Someone to review my work as I go	27%
Didn’t know contributions were open to anyone	22%
Help setting up a development build	20%

[ghutchis]

(This is a placeholder - I’ll finish later tonight - time to go teach)

[ghutchis]

Tasks

Not surprisingly, a lot of building, editing and visualization. Less on the analysis and preparing and submitting calculations.

Here’s what people say we do best right now. (Hmm, guess we should work some on analysis.)

[ghutchis]

What We Do Well

Most-repeated: simplicity and ease of use, then free/open-source, visualization quality, molecule building, and input generation. [ Honestly, I think the visualization quality is much higher than in v1.2 ]

Several people framed Avogadro as the successor to other tools — Molden, Jmol, and Chem3D were all named. One respondent listed the things that have been fixed recently — orbital visualization, output parsing, trajectory and frequency analysis — and credited the developer community for being responsive.

[ghutchis]

What most needs improvement

Ranked by how many people raised each.

Manipulation and selection (7 and 8 respondents)

The biggest grouping, and remarkably consistent in wording. The manipulate/rotate tool is described as unintuitive and clunky; selection is hard to control; deleting hydrogens and bonds doesn’t behave the way it did in 1.2. Chemcraft and Avogadro 1.2 are the repeated comparisons.

Rotating parts of the molecule is not intuitive

The manipulation tool for rotating […] is more hassle than it is worth

[ Suggestions and specific feedback is very welcome here, since we largely just ported code from Avogadro v1.2 ]

One crystallographer can’t reliably select specific molecules out of a set of dimers. Another notes that atoms which appear close together on screen can be far apart in depth, which makes bond creation error-prone.

Constraints and interactive optimization (9 respondents)

Freezing atoms or regions during optimization, fixing a hydrogen-bond angle, and optimizing while editing rather than waiting. One person described the use case precisely: steering a structure mid-optimization to land on the right stereoisomer, which worked in Avogadro 1.

[ Some of this already exists — see below ]

Periodic and solid-state (9 respondents)

A coherent cluster: CIF files not parsing correctly (one person has moved to Mercury and CrystalMaker), space groups and asymmetric units, the repeated space-group prompt when building crystals, surfaces and interfaces and adsorbates, amorphous cells for polymer work, and periodic optimization with ASE machine-learning potentials. OVITO and VESTA are named as the alternatives people reach for.

[ Again, specific suggestions are welcome - I know @Thomas has made a few ]

File parsing — ORCA particularly (6 respondents)

ORCA output parsing is the most-repeated concrete parsing comment in the survey. The newest response is the most specific and the most concerning:

Avogadro v1.2 could visualize and analyze DFT optimization output file from ORCA v6.1. The Avogadro v2 can do neither, just error messages.

I’d very much like a failing file. If you have an ORCA output that 1.2 reads and 2.x doesn’t, please attach it to a reply, open a thread or open an issue.

Also raised: PDB handling — atom and residue names, numbering, chirality, hydrogens landing in a new residue instead of the parent, and residue sequences breaking on edit.

Crashes (6 respondents)

too many random ways to crash Avogadro

Not much more detail than that, which is the problem. If you’ve experienced a crash, please let us know - we can’t fix it otherwise.

That said, the new crashpad build for Windows will be open for testing soon.

Input generators (6 respondents)

A few comments: “functionals, basis sets and keywords are thin for everything except Gaussian”. Also requests for counterpoise/BSSE in the ORCA generator, and easily customisable FT-IR and UV spectra plotted directly from output files.

[ There’s an open thread for BSSE - but if you have suggestions, please open a thread or post a request with the avogenerators.

Smaller but recurring

Publication-quality image export; exporting atomic charges; saving structures displaced along a normal mode; an in-app format converter; RMSD-based atom index mapping between structures; NCI visualization and Windows enterprise installer problems.

[ More feedback is particularly welcome - for example how do you want to export atomic charges? ]

[ghutchis]

Things people asked for that Avogadro 2 already does

This is the finding that I most want to discuss…

Several of the most-requested “missing features” already exist. In some cases the same capability appears in the survey as both a feature request and a usability complaint — meaning a chunk of our worst feedback is about functionality we shipped.

Requested in the survey	Actually available
Set a bond length, angle, or dihedral to a numeric value
Constrained optimization — e.g. fixing a hydrogen-bond angle
Freeze specific atoms or regions during optimization
Conformer search	also via plugins
Distort along a vibrational mode and save the structure
Generate a structure from a name or SMILES

Two respondents asked for numeric dihedral setting citing Chemcraft, and two separate complaints about the manipulate tool are essentially describing a feature that exists.

So this is a documentation and discoverability problem. The plan:

1. More task-oriented “how do I…” tutorials in the docs, keyed to the exact phrasings people used in this survey.

2. In-application discoverability work — menu placement, tooltips, disabled-state affordances. (And looking into integrating the manual docs into a help system.)

This is also the single easiest entry point for anyone who wants to help but doesn’t write C++. If you know how to do one of the things in that table, writing it up is a genuine contribution. You can just send us the text or write it here on the forum and we’ll post it to the website.

[ghutchis]

Contributing

1. We need to surface good first issue and help wanted tasks and ideally each would point to parts of the code.

2. Small contributions are welcome. Explicitly. A typo fix, one compiler warning, one documentation page. That is a real contribution and it’s honestly a big help. Avogadro is a huge community and every bit has made a better tool.

3. Contributions are open to anyone. No invitation needed, no affiliation needed, no permission needed.

On Python

Of 49 respondents, 44 code at least a little. Python is near-universal (37); C++ is 16. Our largest pool of willing contributors - so we need to make it easier to add and improve plugins.

The packages people asked for most: RDKit (7) and ASE (6), then pymatgen, Psi4, and Open Babel. [ Honestly, this is another discovery challenge - we already have plugins using RDKit, ASE, pymatgen and Psi4 ]

If you write Python and want to help, plugin work is the place to start and I’d rather you asked a “dumb” question here than stay silent.

On credit

29% said academic credit would make a difference, and one long-standing contributor was candid that they’d spend more time on the project with some form of support. I’d welcome thoughts on it - particularly in terms of fundraising.

[ghutchis]

Financial support

39% said no, which is completely fine. But 4 people said “I would, but I don’t know how” and 6 said their department or institution might pay. So we need to make paths more visible - and I’m working on a few things, including with Pitt (i.e., in the US, donations to a university aren’t taxable).

Several respondents apologized for not being able to contribute financially — one wrote that their country is severely underfunded, another simply “ima poor student :(”. Please don’t apologize. Avogadro is free and will always stay free. If we add a donation link it will be clearly optional, and using the software without paying for it is exactly what it’s for. Filing a feature idea or good bug report is worth more to us than a small donation anyway.

One person specifically asked for one-time or voluntary donation models rather than subscriptions. Yep, that’s already possible.

[ghutchis]

How you can help right now

• Have an ORCA output file that 1.2 reads and 2.x won’t? (Or some other file that doesn’t work.) Post it. That’s the highest-value thing you could do today.

• Went back to 1.2 for a specific reason? Tell us what it was — there’s a good chance it’s fixed.

• Know how to do something from the “already exists” table in above Write it up. That’s a real contribution.

• Want to write code but don’t know where to start? Reply here, say roughly what you’re comfortable with, and we’ll point you at something concrete with a file path attached.

Thanks again to everyone who responded — and particularly to the people who wrote long, critical answers. Those are extremely helpful and well worth the time you invested.

[matterhorn103]

Thanks for the detailed analysis. Some very useful stuff there!

Clearly some of it is just about waiting for newer versions of the program to diffuse out to people, but that’s fine, the Linux distros will catch up gradually.

Particularly thinking about the focus of my own efforts in light of the results:

• It’s good to have confirmation of the high Python literacy. Like you say, really tight documentation and tutorials around plugins, so that people genuinely feel like it’s easy to do and that they have all the information available to them, is clearly a key thing to prioritize. I know from my own experience what it’s like to write something based on good documentation for an API or framework vs bad, and it makes a world of difference. A definite short-term focus area for me.

• Based on the ranking of “analysis” and the requests re: plotting, ORCA parsing etc., I’d say people are really keen to be better able to use Avogadro to “look at the results of calculations” in general. There’s a lot that can be done there.

Thinking of both those aspects together, it occurs to me that one of the most common things chemists do with Python is plotting data, and that leads me to ponder an idea … I’ll put up a thread about it later, I think.