xAI's 2026 flagship. Frontier-competitive reasoning with real-time X data, at aggressive pricing.
Grok 4.20 is xAI's 2026 flagship and its bid to compete on price as well as personality. Where the legacy Grok 4 launched at $3/$15 — premium-tier pricing, and it remains available at that rate — 4.20 lands at $2 input and $6 output per 1M tokens, undercutting every other flagship on our API table. It's competitive with frontier models on benchmarks, and the X (Twitter) integration remains the genuinely unique feature: real-time access to the timeline that no other provider has.
The model voice is also distinct — less filtered, more conversational — which is a feature or a bug depending on your product. For consumer apps that want personality, it's a draw; for enterprise document work, the Claude and GPT families remain the safer defaults.
Two budget siblings share the family. Grok 4.1 Fast ($0.20/$0.50 per 1M) is the speed play with a standout spec: a 2M token context window, the longest on our entire table — twice GPT-5.5's and Fable 5's 1M. If your problem is "read something enormous, cheaply", nothing else comes close on price. Grok Code Fast 1 ($0.20/$1.50) targets code tasks at the same input rate. Both trade reasoning depth for throughput — use 4.20 when the answer needs to be smart, the Fast models when it needs to be fast and cheap.
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Grok 4.20 | $2 | $6 | 256K |
| Grok 4 (legacy) | $3 | $15 | 256K |
| Grok 4.1 Fast | $0.20 | $0.50 | 2M |
| Grok Code Fast 1 | $0.20 | $1.50 | 256K |
| Gemini 3.1 Pro | $2 | $12 | 1M |
| Claude Sonnet 4.6 | $3 | $15 | 200K (1M β) |
At $2/$6, Grok 4.20 matches Gemini 3.1 Pro on input and halves it on output, while Sonnet 4.6 costs 2.5× as much for output. The honest trade: those rivals have stronger track records on careful long-form work and bigger ecosystems. Grok wins on price, real-time data, and voice; it has the most to prove on reliability-critical workloads.