How UPI payments work¶
A short explainer: how one UPI payment moves money from your bank to a shop's bank in seconds, through four parties and five steps. The narration is by Gnani.ai's Avinash voice.
Made by Claude Code with manimvox
This video was made by Claude Code with Sonnet 5.5, from one prompt and with no further help. It took 3 minutes and cost $0.40. The agent found the sources, wrote the facts, the script and the episode, drafted it, read its own checks and rendered it.
The prompt¶
"Make a short video, about a minute, explaining how UPI payments work. You may search and fetch the web for sources. Use the default voice."
The same episode was then rendered with three voices (Sarvam AI, Gnani.ai and ElevenLabs) to compare them. The leaderboard below shows the results, and the benchmark report has all three videos and the full review.
Leaderboard¶
How the voices, agent harnesses and models tried on this example compare, from 1 benchmark run. The methodology explains every measurement.
Voice¶
Ranked by the listener's order within each run, then word error rate, then crest factor (higher is less compressed, cleaner audio). Each voice read the same lines as the others in its run.
| # | Voice | Word errors | Words a minute | Crest factor | Time per line | Cost | Listener | Run |
|---|---|---|---|---|---|---|---|---|
| 1 | Gnani.ai, Timbre v3.0, Avinashgnani:timbre-v3.0/Avinash@en-IN |
0.6% | 171 | 20.2 dB | 3.97 s | $0.029 | the most realistic; clear, but slow | 2026-10-23 |
| 2 | ElevenLabs, Eleven v3, Brianelevenlabs:eleven_v3/nPczCjzI2devNBz1zQrb |
0.0% | 169 | 18.0 dB | 2.96 s | $0.18 | the best pacing; less noise and distortion than Sarvam; sounds AI-generated | 2026-10-23 |
| 3 | Sarvam AI, Bulbul v3, shubhsarvam:shubh |
2.2% | 215 | 13.5 dB | 1.36 s | $0.029 | sounds AI-generated; not clear, background noise; good pace for a short video | 2026-10-23 |
Harness¶
Ranked by result, then cost, then time.
| # | Harness | Model | Example | Result | Time | Turns | Cost | Run |
|---|---|---|---|---|---|---|---|---|
| 1 | Claude Code 2.1.294 | claude-sonnet-5-5 |
How UPI payments work | video made | 3 min 3 s | 21 | $0.40 | 2026-10-23 |
Model¶
Every run with each model, ranked by the share that made a video, then mean cost.
| # | Model | Runs | Videos made | Mean cost | Mean time | Harnesses |
|---|---|---|---|---|---|---|
| 1 | claude-sonnet-5-5 |
1 | 100% | $0.40 | 3 min 3 s | Claude Code |
Source¶
The project is in benchmarks/runs/2026-10-23-upi-voices/project: the episode, its script, the facts and the subtitles. To render it yourself with this voice:
uv tool install "manimvox[kokoro,cloud]" # or: pip install "manimvox[kokoro,cloud]"
export GNANI_API_KEY=...
manimvox render ep01_upi --voice gnani:timbre-v3.0/Avinash@en-IN
Credits¶
- Facts: Wikipedia, Unified Payments Interface, and two explainers listed in the project's
FACTS.md. - Voice: Gnani.ai, Timbre v3.0, Avinash.
- Made by: Claude Code (Sonnet 5.5) with manimvox.