Grok 4.3
2026-04-30 · Publisher : xAI · Domain : Artificial Intelligence
xAI rolls out Grok 4.3, a model with built-in reasoning and a 1 M-token context, at a sharply reduced price (1.25 USD input, 2.50 USD output per million tokens).
What changes
- Beta that appeared on April 17, 2026 in the grok.com selector (reserved for SuperGrok Heavy), full API rollout around April 30 - May 1.
- New pre-training at a scale equivalent to Grok 4.20, improved architecture.
- Native video input, document generation (PDF, spreadsheets, slides) and better tool calling (third-party sources).
- Price cut of about 40% on input and 60% on output compared with the previous generation (VentureBeat).
### Specifications - API identifier: grok-4.3. - Context: 1 M tokens. - Knowledge cutoff: December 2025 (Wikipedia). - Availability: xAI API, Grok app. ### Pricing | Item | Input | Output | |---|---|---| | Standard (under 200k tokens) | 1.25 USD | 2.50 USD | | Cache | 0.20 USD | - | | Tier of 200k tokens and above | 2.50 USD | 5 USD | The long-context tier shown by third-party sources (4 / 12 USD) does not match the current docs.x.ai. ### Benchmarks No official figure in the sources reviewed. VentureBeat cites 79.3% on CaseLaw v2 (third-party measurement). ### Key takeaways Very economical mid-range model, still listed in docs.x.ai in late September 2026.
Links
Tags : LLM, xAI, Grok, Raisonnement, Long contexte