Compare

Grok 4.20 Non-Reasoning vs Grok 4.5

Pricing per million tokens, context window, capabilities — pulled from each provider's public docs. All 2 are available via the same AIgateway OpenAI-compatible endpoint; flip the model string to switch.

Search2/4
Grok 4.20 Non-Reasoning
xai/grok-4.20-0309-non-reasoning
Grok 4.5
xai/grok-4.5
Provider
xAI
xAI
Family
Grok 4
Grok 4
Modality
text
text
Context window
2,000,000 tok
500,000 tok
Max output
32,768 tok
32,768 tok
Released
2026-05-22
2026-07-21
License
Proprietary
Proprietary
Input price
$2.00 /1M
$1.00 /1M
Output price
$6.00 /1M
$1.00 /1M
Tools
Streaming
yes
yes
Vision
yes
yes
JSON mode
Reasoning
Prompt caching
Batch API
Try it
Open in playground →
Open in playground →
Grok 4.20 Non-Reasoning
xai/grok-4.20-0309-non-reasoning
Full spec →

xAI's Grok 4.20 non-reasoning model. Skips the thinking trace for fast, single-pass responses while keeping the same training as the reasoning variant.

Strengths
  • General-purpose chat
  • Long context
  • Tool use
Use cases
ChatbotsContent generationAgentic workflows
Grok 4.5
xai/grok-4.5
Full spec →

xAI's Grok 4.5, a frontier model built for coding, agentic tasks, and knowledge work. Accepts text and image inputs, and supports function calling, structured outputs, and configurable reasoning effort (low, medium, high).

Strengths
  • General-purpose chat
  • Long context
  • Tool use
Use cases
ChatbotsContent generationAgentic workflows

Compare with another

Claude Opus 5 vs Grok 4.5
anthropic/claude-opus-5 · xai/grok-4.5
Grok 4.5 vs Kimi K3
xai/grok-4.5 · moonshot/kimi-k3
Grok 4.5 vs Gemini 3.6 Flash
xai/grok-4.5 · google/gemini-3.6-flash
Grok 4.5 vs GPT-5.6 Sol
xai/grok-4.5 · openai/gpt-5.6-sol
Grok 4.5 vs Claude Fable 5
xai/grok-4.5 · anthropic/claude-fable-5
Grok 4.5 vs DeepSeek V4 Pro
xai/grok-4.5 · deepseek/deepseek-v4-pro
Grok 4.5 vs Gemini 3.1 Pro
xai/grok-4.5 · google/gemini-3.1-pro
Grok 4.5 vs MiniMax M3
xai/grok-4.5 · minimax/m3
Grok 4.5 vs Kimi-K2.6
xai/grok-4.5 · moonshot/kimi-k2.6
SWITCH BETWEEN THEM

One key, all 2, one line different.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.aigateway.sh/v1",
    api_key="sk-aig-...",
)

# Grok 4.20 Non-Reasoning
client.chat.completions.create(
    model="xai/grok-4.20-0309-non-reasoning",
    messages=[{"role":"user","content":"hello"}],
)

# Grok 4.5
client.chat.completions.create(
    model="xai/grok-4.5",
    messages=[{"role":"user","content":"hello"}],
)
Get an AIgateway keyRun an eval on these →