Data read October 27, 2026
Gradium TTS vs Inworld Realtime TTS-2: TTS API comparison
On P95 latency, Gradium TTS is faster: 104.3 ms for Gradium TTS against 194.7 ms for Inworld Realtime TTS-2.
On Artificial Analysis listener Elo, Inworld Realtime TTS-2 is ahead (1,174 for Gradium TTS, 1,267 for Inworld Realtime TTS-2). List price per million characters: $36 to $57.80 for Gradium TTS, $15 to $25 for Inworld Realtime TTS-2.
Gradium TTS is ranked 1 of 23; Inworld Realtime TTS-2 is ranked 6 of 23. Every comparison on this site uses Gradium TTS as its fixed reference, and no page says either model is better overall.
| Measure | Gradium TTS | Inworld Realtime TTS-2 | Ahead |
|---|---|---|---|
| Rank in the multilingual speed ranking | 1 | 6 | n/a |
| Speed tier | Headroom | Fast enough | n/a |
| Median time to first audio (ms), lower is faster | 53.7 | 149.1 | Gradium TTS |
| P95 time to first audio (ms), lower is faster | 104.3 | 194.7 | Gradium TTS |
| Listener Elo (Artificial Analysis) | 1,174 | 1,267 | Inworld Realtime TTS-2 |
| Customer-service Elo (Artificial Analysis) | 1,241 | 1,311 | Inworld Realtime TTS-2 |
| Naturalness (Speko) | 1,592 | 1,561 | Gradium TTS |
| Naturalness estimate (Cekura, 1 to 5) | 4.51 | 4.52 | Inworld Realtime TTS-2 |
| Reading accuracy (Cekura) | 79.5% | 93.4% | Inworld Realtime TTS-2 |
| Hard passages judged correct (Artificial Analysis) | n/a | 80.7% | n/a |
| Lowest list price per 1M characters (lower is cheaper) | $36 to $57.80 | $15 to $25 | Inworld Realtime TTS-2 |
| Languages tested (Artificial Analysis, Speko) | 4 | 12 | Inworld Realtime TTS-2 |
| Access | Proprietary | Proprietary | n/a |
| Streaming status | Measured | Measured | n/a |
Ahead is computed from the figures only. A latency gap under 10 ms is a tie; listener Elo values whose 95% intervals overlap are a tie. A measure one benchmark does not hold is n/a for that model and is never estimated. See the methodology and each model page for sources.
Frequently asked questions
- Is Gradium TTS faster than Inworld Realtime TTS-2?
- On P95 latency, Gradium TTS is faster: 104.3 ms for Gradium TTS against 194.7 ms for Inworld Realtime TTS-2.
- How do Gradium TTS and Inworld Realtime TTS-2 compare on voice quality?
- On Artificial Analysis listener Elo, Inworld Realtime TTS-2 is ahead (1,174 for Gradium TTS, 1,267 for Inworld Realtime TTS-2).
- Which is cheaper, Gradium TTS or Inworld Realtime TTS-2?
- List price per million characters: $36 to $57.80 for Gradium TTS, $15 to $25 for Inworld Realtime TTS-2.