TTS API comparison 2026
Gradium TTS vs Inworld Realtime TTS-2 Flash Gradium vs Inworld · text-to-speech for voice agents
Gradium TTS is in the Headroom speed tier (P95 under 120 ms); Inworld Realtime TTS-2 Flash is in the Headroom speed tier (P95 under 120 ms). Below, each measure side by side with its source; a model leads on a measure only beyond the tie rule. No combined score, no overall winner.
Figures as of October 27, 2026 · Coval, Speko, Artificial Analysis, Cekura
Side by side
| Measure | Gradium TTS | Inworld Realtime TTS-2 Flash |
|---|---|---|
| Speed tierCoval P95, 30 days | Headroom | Headroom |
| Median time to first audioCoval, 30-day window | 54 mstie | 58 mstie |
| P95 time to first audio (slow turns)Coval, 30-day window | 104 mstie | 101 mstie |
| Time to first audio, board chartSpeko, US East | 55 mstie | 50 mstie |
| Time to first audio, full sentence (median)Cekura | 76 ms | 46 ms |
| Time to first audio, text streamed word by word (median)Cekura | -64 ms | 45 ms |
| Naturalness EloSpeko | 1,592 | 1,501 |
| Listener Elo (95% interval)Artificial Analysis | 1,174 ±13 | 1,218 ±13 |
| Numbers and dates (0 to 1)Speko | 0.27 | 0.58 |
| Hard passages pronounced correctly (95% interval)Artificial Analysis, human reviewers | not tested | 77.8% (74.8 to 80.7) |
| Reading accuracy on hard textCekura, transcriber-scored | 79.5% | 93.2% |
| List price per 1M charactersCoval, Speko, Artificial Analysis, Cekura | $36 to $57.8 | $10.4 to $15 |
In bold: the model ahead on that measure. “tie”: latencies within 10 ms, or Elo scores whose 95% intervals overlap.
Where Gradium TTS is ahead
- →Time to first audio, text streamed word by word (median): -64 ms vs 45 ms (Cekura)
- →Naturalness Elo: 1,592 vs 1,501 (Speko)
Where Inworld Realtime TTS-2 Flash is ahead
- →Time to first audio, full sentence (median): 46 ms vs 76 ms (Cekura)
- →Listener Elo (95% interval): 1,218 ±13 vs 1,174 ±13 (Artificial Analysis)
- →Numbers and dates (0 to 1): 0.58 vs 0.27 (Speko)
- →Reading accuracy on hard text: 93.2% vs 79.5% (Cekura, transcriber-scored)
- →List price per 1M characters: $10.4 to $15 vs $36 to $57.8 (Coval, Speko, Artificial Analysis, Cekura)
Frequently asked questions
- What is the difference between Gradium TTS and Inworld Realtime TTS-2 Flash?
- Gradium TTS (Gradium) is in the Headroom speed tier (P95 under 120 ms); Inworld Realtime TTS-2 Flash (Inworld) is in the Headroom speed tier (P95 under 120 ms). Gradium TTS leads on time to first audio, text streamed word by word (median) and naturalness Elo; Inworld Realtime TTS-2 Flash leads on time to first audio, full sentence (median), listener Elo (95% interval), numbers and dates (0 to 1), reading accuracy on hard text and list price per 1M characters. They are a tie on median time to first audio, P95 time to first audio (slow turns) and time to first audio, board chart. No combined score is computed.
- Which is faster for real-time voice agents, Gradium TTS or Inworld Realtime TTS-2 Flash?
- On Coval's 30-day window as of October 27, 2026, the median time to first audio is 54 ms for Gradium TTS and 58 ms for Inworld Realtime TTS-2 Flash (a tie); the P95, the slow turn, is 104 ms and 101 ms (a tie). Gaps under 10 ms are ties.
- Which sounds more natural, Gradium TTS or Inworld Realtime TTS-2 Flash?
- Speko naturalness: 1,592 for Gradium TTS, 1,501 for Inworld Realtime TTS-2 Flash. Artificial Analysis listener Elo: 1,174 ±13 and 1,218 ±13. The two sources use different listeners and can disagree.
- How are these two text-to-speech APIs compared?
- Speed first, on Coval's median and P95 time to first audio; then voice quality (Speko, Artificial Analysis), accuracy on numbers and dates (Speko) and list price. Latencies within 10 ms and Elo scores whose intervals overlap are ties. Scores from different benchmarks are never added into one number.
- Can these numbers be checked?
- Yes. Each model page lists every measurement with its source and the date of the reading, and links to the benchmark page it comes from. The rules are on the methodology page.