Currently listed through these providers:
Model details
Grok 4.20 (Non-Reasoning)
Grok 4.20 Non-Reasoning is the streamlined sibling within xAI's Grok 4.20 family, purpose-built for fast agentic tool calling rather than extended deliberation. Oracle's documentation describes the full Grok 4.20 as offering both reasoning and non-reasoning variants, with the non-reasoning cut optimized to skip chain-of-thought overhead entirely. This design trades deep step-by-step reasoning for speed, making it suited for production pipelines and automation scenarios where rapid, reliable tool-use matters more than verbose internal reasoning. xAI positions the model around the claim of the lowest hallucination rate on the market, paired with strict prompt adherence that keeps responses aligned to user intent without wandering.
The model carries an Intelligence Index score of 29.0 and a Coding Index of 22.0 according to benchmarks referenced across multiple sources, reflecting its more specialized focus compared to the full reasoning variant. CloudPrice notes it's among roughly 505 LLMs tracked, ranking around the 156th tier with output speeds reaching 89 tokens per second and first-token latency around 0.49 seconds, highlighting the speed advantages this non-reasoning approach delivers. The design philosophy leans into agentic deployments where developers want trustworthy, fast responses without the latency cost of chain-of-thought chains, making Grok 4.20 Non-Reasoning a practical choice for developers building automation, structured tool calls, or high-throughput AI workflows that demand both speed and reliability.
Quick Info
Powered by- Provider
- DevPass (LLM Gateway)
- Model key
- grok-4-20-beta-0309-non-reasoning
- Release date
- Mar 9, 2026
- Last updated
- Mar 9, 2026
- Input modalities
- Output modalities
- Capabilities
Cost
- Input token cost
- $2.00
- Output token cost
- $6.00
Limits
- Output tokens
- 30,000 tokens
- Context window
- 2,000,000 tokens
Transparent token rates
Compare Grok 4.20 (Non-Reasoning) pricing
Rates are shown per one million tokens. Combined means one million input plus one million output tokens.
Latest news about Grok 4.20 (Non-Reasoning)
No articles yet. Fetch the latest news to show it here.