OpenAI
GPT-5.3 Codex Spark
OpenAI's ultra-low-latency coding model released in 2026, built for real-time coding collaboration and rapid iteration with caching support
Context Length128K tokens
Model TypeUltra-Low-Latency Code Model (Small)
Key FeaturesReal-time coding, rapid iteration, targeted code edits
Pricing & Specs
💰 Pricing
Input$1.75 / M tokens
Output$14.0 / M tokens
Cache Hit$0.175 / M tokens
⚙️ Specs
Context Length128K tokens
Model TypeUltra-Low-Latency Code Model (Small)
Key FeaturesReal-time coding, rapid iteration, targeted code edits
Generation Speed1000+ tokens/s (official demo)
ModalitiesText input, text output
AvailabilityResearch Preview (February 12, 2026)
API CompatibilityOpenAI API, Codex CLI
API Examples
Python (OpenAI SDK)
from openai import OpenAI
client = OpenAI(
api_key="your-api-key",
base_url="https://api.xairouter.com/v1"
)
response = client.chat.completions.create(
model="gpt-5.3-codex-spark",
messages=[
{"role": "user", "content": "Write a quicksort function in Python"}
]
)
print(response.choices[0].message.content)cURL (OpenAI API)
curl https://api.xairouter.com/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"model": "gpt-5.3-codex-spark",
"messages": [
{"role": "user", "content": "Write a quicksort function in Python"}
]
}'Developer Assist
# Configure ~/.codex/config.toml
cat > ~/.codex/config.toml << 'EOF'
model_provider = "xai"
model = "gpt-5.3-codex-spark"
approval_policy = "never"
sandbox_mode = "danger-full-access"
network_access = true
preferred_auth_method = "apikey"
[shell_environment_policy]
inherit = "all"
ignore_default_excludes = false
[model_providers.xai]
name = "xai"
base_url = "https://api.xairouter.com"
wire_api = "responses"
requires_openai_auth = false
env_key = "OPENAI_API_KEY"
web_search = true
EOF
# Set environment variable (add to ~/.bashrc or ~/.zshrc)
export OPENAI_API_KEY="sk-Xvs..."
# Launch Codex CLI
codex