MoonshotAI Kimi K2 Thinking vs SpaceXAI Grok 4.3
Side by side from live data: prices, limits, who serves each model and how reliably, and how they do on our chess arena.
In short
- Kimi K2 Thinking is 1.5× cheaper at a typical mix of 3 input tokens per output token.
- Grok 4.3 accepts longer prompts: 1M vs 262K tokens.
- Kimi K2 Thinking's best provider was more reliable over 24 h: 99.88% vs 99.17%.
- Grok 4.3 is the newer model (released 2026-04-30).
| Kimi K2 Thinking | Grok 4.3 | |
|---|---|---|
| Vendor | MoonshotAI | SpaceXAI |
| Released | 2025-11-06 | 2026-04-30 |
| Input / 1M | $0.60 | $1.25 |
| Output / 1M | $2.50 | $2.50 |
| Cached input / 1M | $0.15 | $0.20 |
| Blended / 1M (3:1) | $1.07 | $1.56 |
| Context | 262K | 1M |
| Max output | 98K | 900K |
| Accepts | text | text, image, file |
| Reasoning | Yes | Yes |
| Providers on OpenRouter | 1 | 4 |
| Cheapest provider, output | $2.50 | $2.50 |
| Best provider uptime, 24 h | 99.88% | 99.17% |
| Chess arena rating | not playing | 1693 ±85 |
| Vendor API status | Operational | Operational |
Cost calculator
Kimi K2 Thinking
$73.50 per month$2.45 per day
Grok 4.3
$113 per month$3.75 per day
Kimi K2 Thinking saves $39.00 a month on this workload.
At list prices per 1M tokens; prompt caching and batch discounts are not included.
Compare Kimi K2 Thinking with