Alibaba vs 
SpaceXAI
Verdict
- Their flagships are level: Qwen3 8 66.5 vs Grok 4.6 66.0.
- Alibaba has the better model in 4 of 4 price bands.
- SpaceXAI's fastest ranked model (Grok 4.20 Multi-Agent, 304 tok/s) beats Alibaba's (Qwen3.5 Omni Flash, 249 tok/s).
- Alibaba publishes more open-weights models (53 vs 1).
- Alibaba ships more often: one model every 7 days against every 42.
Best model per price band
Each provider’s highest-index model below the price ceiling. Purple marks the winner of the band.
| Price band | Alibaba | SpaceXAI |
|---|---|---|
| Under $0.50/M | Qwen3.8-Flash-Next 61.5$0.230/M | – |
| Under $2/M | Qwen3.8-Flash-Next 61.5$0.230/M | Grok 4.20 thinking 59.9$1.56/M |
| Under $10/M | Qwen3 8 max 66.5$2.67/M | Grok 4.6 medium 66.0$3.00/M |
| Any price | Qwen3 8 max 66.5$2.67/M | Grok 4.6 medium 66.0$3.00/M |
Line-up
| Metric | Alibaba | SpaceXAI |
|---|---|---|
| Flagship index | 66.5 | 66.0 |
| Ranked models | 55 | 16 |
| Open-weights models | 53 | 1 |
| Cheapest ranked, $/M | $0.060 | $1.25 |
| Fastest ranked, tok/s | 249 | 304 |
| Release cadence, days | 7 | 42 |
| Last release | 1 Sept 2026 | 21 Sept 2026 |
| Next expected | 5 Sept 2026 – 10 Sept 2026 | 16 Oct 2026 – 18 Nov 2026 |