Model choice without rewrites
Evaluate quality and unit economics across coding models while your client contract stays consistent.
AI CODING AGENTS
Build coding agents that can reason, use tools and ship.
Give your developer product the model choice, tool interface and production controls it needs—without rebuilding the integration every time a stronger coding model arrives.
PRODUCTION ENDPOINT
POST /v1/chat/completions200K+LONG CONTEXT
SSESTREAMING
TOOLSFUNCTION CALLS
Production advantageMove faster without
Move faster without
shrinking your options.
AIx keeps identity, integration and measured billing consistent while your workload uses the model best suited to the job.
Tool-native execution
Send function schemas, parse tool calls and return execution results using familiar OpenAI-compatible types.
Responsive agent UX
Stream partial output to terminals, editors and web interfaces while the model continues reasoning.
Recommended modelsStart with a model
Start with a model
matched to the work.
Use these as practical starting points, then evaluate quality, latency and unit economics against your own traffic.
Reference architectureFrom request
From request
to production.
- 01Retrieve the right context
Search the repository, rank relevant files and keep the prompt inside the selected context window.
- 02Let the model plan and call tools
Expose safe functions for file reads, tests, diffs and targeted edits.
- 03Validate before applying
Run tests and policy checks, then present or apply the final patch through your own permission layer.
QUICKSTART
POST /v1/chat/completionsfrom openai import OpenAI
client = OpenAI(
base_url="https://api.aix.theaimart.co/v1",
api_key="$AIX_KEY",
)
stream = client.chat.completions.create(
model="deepseek-ai/DeepSeek-V3.2",
messages=[{"role": "user", "content": "Review this patch"}],
tools=tools,
stream=True,
)FAQ
Questions, answered.
Which model should I use for a coding agent?+
Start with a quality-focused coding model for difficult repository work, then test an efficient open model for high-volume background tasks. AIx keeps both behind the same client contract.
Does AIx support tool calling for coding agents?+
Yes. Eligible models accept OpenAI-compatible tools and tool_choice fields and return standard tool_calls.
Can coding-agent responses stream?+
Yes. Set stream to true and consume standard server-sent events until the final DONE event.
Ship the first request today.
One key. One API contract. The AIx model network behind it.