Grok 4.3
Previous cheap-frontier Grok — superseded as flagship by Grok 4.6 on 12 August 2026.
Read the full Grok 4.3 analysisContext
1M
Max output
—
Input /1M
$1.25
Output /1M
$2.50
Live pricing via OpenRouter
Best for
- High-volume cost-sensitive inference (classification, extraction, summarization)
- Latency-sensitive agentic tool loops
- Native video-input and voice-cloning workflows
Watch out
Context window dropped 2M → 1M; per-token price doubles past 200K tokens in a single request; reasoning is always-on so trivial calls cost slightly more.
For creators. Native video input removes a transcription step for footage analysis; Custom Voices enables narration and conversational agents at a flat per-minute rate.
Benchmarks
| aa intelligence index | 53 |
| gdpval aa | 1500 |
| output tokens per sec | 181.3 |
Capabilities
- Always-on reasoning by default
- 1M-token context window
- Native video input (first for xAI API)
- Custom Voices voice-cloning suite
- Agentic tool use (web search, Python sandbox, PDF/Excel/PPTX generation)
- Highest output throughput in its price tier (~181 tok/s)
Compare Grok 4.3
Grok 4.6 vs Grok 4.3
Use 4.6 as the current xAI flagship for long agent loops. Keep 4.3 when the $1.25/$2.50 band still fits. Do not treat published AA/vendor tables as SIS Model Arena winners.
Claude Fable 5 vs Grok 4.3
Different products. Fable 5 is the agentic-coding ceiling; Grok 4.3 is the cheapest credible frontier intelligence with the fastest throughput in its class. The 20× output-price gap means most stacks should run both — at different tiers.
Grok 4.3 vs GPT-5.5
GPT-5.5 is clearly the stronger model; Grok 4.3 delivers a large share of the capability at roughly a fifth of the price — the budget-frontier default.