EmoteBench · Keyframe Labs Research

A new standard for expressive avatars

Avatar benchmarks measure lip sync, latency, and preference. EmoteBench measures natural, emotional expression. When the sentiment is grave we expect to see concern; when it brightens, we expect the avatar to brighten with it.

EmoteBench is a blinded human study of whether real-time avatars do that. 250 participants made 19,500 judgments across five avatar APIs, rating whether each avatar expressed the right emotion, looked natural, and changed when prompted. Keyframe Labs scored highest on every measure, against every competitor.

Expression

2.2×the closest competitor

Muted clips where participants named the avatar's emotion

Keyframe Labs81%
Tavus37%

95% CI 76–85%

Naturalness

1.9×the closest competitor

Ratings of very or completely natural

Keyframe Labs61%
Anam32%

95% CI 54–67%

Control

4×the closest competitor

Prompted emotion switches where participants named both emotions

Keyframe Labs53%
Tavus13%

95% CI 41–64%

Expression

Clearer emotions, clearer intent.

Shown with an avatar expressing an emotion, participants picked which of seven emotions it showed. For Keyframe Labs' avatar, they picked the intended emotion 81% of the time. Anam, Tavus, and LemonSlice landed between 32% and 37%, where guessing scores 14%. No other avatar was more emotionally readable for 239 of the 250 participants.

Keyframe Labs
81%
Tavus
37%
Anam
34%
LemonSlice
32%

Share of muted clips where participants chose the intended emotion. About 2,250 ratings per system; the dashed line is chance.

By emotion
EmotionKeyframe LabsAnamTavusLemonSlice
Anger97%22%41%67%
Happiness95%49%94%41%
Neutral87%77%30%76%
Sadness84%16%45%8%
Fear79%2%12%4%
Surprise69%25%30%15%
Disgust58%49%13%10%

Keyframe Labs has the highest share on all seven emotions, Ekman's six basic emotions plus neutral, though Tavus is within a point on happiness. On fear, the three other systems score between 2% and 12%.

Naturalness

More expressive, and more natural.

With the sound on, participants also rated Keyframe Labs' avatar both more natural and clearer in expressing the emotion than every other system. Even the lowest-rated of Keyframe Labs' three avatars outranked every competitor's best.

Looked natural

Keyframe Labs
61%24%15%
Anam
32%31%36%
Tavus
22%26%52%
LemonSlice
26%28%46%
HeyGen
20%19%61%
very or completelysomewhatslightly or not at all

Showed the emotion clearly

Keyframe Labs
79%15%
Anam
29%22%50%
Tavus
27%23%50%
LemonSlice
23%19%58%
HeyGen
32%25%43%
clearly or very clearlymoderatelyslightly or not at all

Share of ratings at the top, middle, and bottom of a five-point scale. About 1,300 ratings per system; HeyGen, which has no emotion control, was rated on neutral and happy lines only.

Naturalness by avatar
LyraKeyframe Labs
3.81
CeliaKeyframe Labs
3.62
CaspianKeyframe Labs
3.42
MiaAnam
3.03
HanaAnam
2.96
Woman 1LemonSlice
2.94
LucyTavus
2.77
Woman 2LemonSlice
2.77
SimonAnam
2.75
HannahHeyGen
2.69
AnnaTavus
2.51
Man 1LemonSlice
2.40
EmilyHeyGen
2.36
DanielTavus
2.34
DanielHeyGen
2.00

Mean naturalness, 1 to 5, for all fifteen avatars. Keyframe Labs' lowest, Caspian at 3.42, sits above the best from any other provider, Anam's Mia at 3.03.

Control

Precise emotional changes.

Participants watched muted clips in which each avatar was prompted to switch emotion mid-sentence, then named the emotion before and after. They named both correctly for Keyframe Labs' avatar 53% of the time, against 13% for Tavus, 5% for LemonSlice, and 3% for Anam. Prompted to hold instead, Keyframe Labs' avatar kept its expression 86% of the time, while Tavus, LemonSlice, and Anam drifted off theirs on 28% to 41% of clips. For those three, participants could barely tell a prompted hold from a prompted switch.

Keyframe Labs
53%
14%
Tavus
13%
32%
LemonSlice
5%
28%
Anam
3%
41%
saw the prompted switch, named both emotionsreported a switch on a clip prompted to hold

Solid: share of prompted switches where participants saw the change and named both emotions; chance is 2%. Hatched: share of control clips, prompted to hold, where participants reported a switch anyway.

Head to head

The near-consensus take: Persona-2.0-Live is better across the board.

Nearly every participant saw every system. For each of the 250 participants we compared their average score for Persona-2.0-Live, Keyframe Labs' model, against their average score for each competitor. On recognizing the emotion, 98% of participants put Persona-2.0-Live ahead; on showing it clearly, 93% to 97%; on looking natural, 77% to 87%. Even the closest comparison, naturalness against Anam, went its way for more than three participants in four. Rating against rating on the same line, it still wins 69% to 87% of the time.

Emotion recognized

vs Anam
98%246 of 250
vs Tavus
98%244 of 250
vs LemonSlice
98%244 of 250
vs HeyGennot tested

Emotion shown clearly

vs Anam
93%232 of 249
vs Tavus
97%240 of 248
vs LemonSlice
97%241 of 249
vs HeyGen
93%228 of 246

Looked natural

vs Anam
77%191 of 249
vs Tavus
87%217 of 248
vs LemonSlice
83%206 of 249
vs HeyGen
85%209 of 246
scored Keyframe Labs higherscored the competitor higher, or the sameone square per participant

One square per participant who saw both systems, 246 to 250 per comparison. Line by line, Keyframe Labs also won 88% to 100% of the lines both systems performed. Every lead's 95% interval excludes zero, both in all sessions and in the half that passed attention checks. HeyGen has no emotion control and was not tested on recognition.

Try it

You be the judge.

Real screens from the study. Pick an answer, then see what the participants chose. Clips play as participants saw them, one system behind each letter. The full report has every breakdown, interval, and example line.

Try it

Behind the Expression result above · No sound. Name the emotion the avatar is showing.

No sound
Clip A · system hidden

Which emotion does the face show?

No sound on this one. Watch the whole clip, then answer.

Which emotion is the face showing?
How sure are you?
A real recognition screen from the study. Each letter is the same line on a different system, in that system's native framing, with no sound. Answer as a participant would, then compare your answer with everyone who rated that clip.

The study

How it was measured

Every system received identical audio for every line, so the only thing that differed between two clips was the avatar. 182 scripted lines across seven emotions, three stock identities per provider, clips captured live from each API and rated blind.

Measured

Late October 2026

Clips captured live from each provider's streaming session, on the versions available at the time.

Sample

250 participants · 19,500 rated clips

An online research panel. One clip per screen, the system shown as a letter, no line seen twice.

Systems

Five real-time avatar APIs

Keyframe Labs persona-2.0-live, Anam cara-4.1-express, Tavus Phoenix-4.5, LemonSlice 2.1, HeyGen LiveAvatar.

Disclosure

Keyframe Labs designed and ran the study

Participants never saw provider names. Every system is reported on every measure it could take.

Read the full report

Code and data: coming soon