ByteDance vs 
OpenAI
Verdict
- OpenAI's flagship leads: GPT-6 Astra at 70.4 vs Dola Seed 2.0 Pro at 58.5.
- OpenAI has the better model in 4 of 4 price bands.
- OpenAI's fastest ranked model (GPT-5.4 Mini, 255 tok/s) beats ByteDance's (Seed-OSS-36B-Instruct, 38 tok/s).
- OpenAI publishes more open-weights models (3 vs 1).
- OpenAI ships more often: one model every 22 days against every 83.
Best model per price band
Each provider’s highest-index model below the price ceiling. Purple marks the winner of the band.
| Price band | ByteDance | OpenAI |
|---|---|---|
| Under $0.50/M | Seed-OSS-36B-Instruct 46.8$0.300/M | GPT-6 Luna max 60.0$0.200/M |
| Under $2/M | Seed-OSS-36B-Instruct 46.8$0.300/M | GPT-6 Luna max 60.0$0.200/M |
| Under $10/M | Seed-OSS-36B-Instruct 46.8$0.300/M | GPT-6.1 Sol max 69.9$4.00/M |
| Any price | Seed-OSS-36B-Instruct 46.8$0.300/M | GPT-6 Astra max 70.4$20.00/M |
Line-up
| Metric | ByteDance | OpenAI |
|---|---|---|
| Flagship index | 58.5 | 70.4 |
| Ranked models | 3 | 44 |
| Open-weights models | 1 | 3 |
| Cheapest ranked, $/M | $0.300 | $0.093 |
| Fastest ranked, tok/s | 38 | 255 |
| Release cadence, days | 83 | 22 |
| Last release | 14 Feb 2026 | 29 Sept 2026 |
| Next expected | 4 Apr 2026 – 10 Jun 2026 | 12 Oct 2026 – 29 Oct 2026 |