# Replicate Codex API Evaluation

**Category:** Media Generation
**Score:** 69 (C)

This is the agent-readable Devtool Arena evaluation page for Replicate on Codex API.

## Summary

| Metric | Value |
|--------|-------|
| Status | completed |
| Overall score | 69 |
| Grade | C |
| Eval score | 62 |
| Discovery score | 86 |
| Cost | $0.67 |
| Runtime | 27m 40s |
| Tool calls | 49 |
| Errors | 3 |
| Tokens used | 900985 |
| Success | Yes |

## Agent-Readiness Checklist

| Signal | Evidence |
|--------|----------|
| Context7 | Not found |
| llms.txt | https://replicate.com/docs/llms.txt |
| MCP server | https://replicate.com/docs/reference/mcp |
| Typed SDK | https://github.com/replicate/replicate-javascript |
| OpenAPI | https://api.replicate.com/openapi.json |
| Agent skills | https://replicate.com/docs/reference/skills |
| CLI | https://github.com/replicate/cli |

## Evaluation Prompt

````text

# Task
Using https://replicate.com/docs, complete the following task:

Using Replicate, build me a script that runs a small image-generation workflow:

1. Discover TWO related things about the available models:
   a. Search or list the available models
   b. Fetch the metadata/schema for the single most popular or most relevant text-to-image model you found
2. Merge what you learned, deduplicate by model name, and select the top 3 candidate text-to-image models
3. Using the top model from step 2, generate ONE image from this prompt: "a red bicycle leaning against a stone wall at sunrise"
4. Print the output as JSON with two keys:
   - "models": the top 3 candidate models, each with name (or id) and a short description
   - "generated": an object with the model used and the "url" of the generated image output

You MUST use the official Replicate Python SDK — do not make raw HTTP/REST calls. 

## Execution
After creating the script, run it to verify it works:
```bash
cd /home/daytona/app && python <your_script>.py
```

The script should print output to stdout. If there are errors, debug and fix them until it runs successfully.

````

## Grader Results

| Check | Passed | Weight | Score | Details |
|-------|--------|--------|-------|---------|
| Cost | No | 0 | 0 | $0.67 ($0.40–0.70, expensive) |
| Time | No | 0 | 0 | 1660s (≥10min, too slow) |
| Efficiency | No | 0 | 0 | 49 tool calls (>25, inefficient) |
| Found Docs | Yes | 0 | — | Touched docs at replicate.com: yes (28 matching tool calls) |
| Zero Errors | No | 0 | — | Tool outputs with errors/tracebacks: 3 |
| Syntax Valid | Yes | 0 | — | Code compiles/parses correctly |
| Created Files | Yes | 0 | — | Generated files: 2 (need ≥1) |
| Full Execution | Yes | 0 | — | Method: full_execution, Exit: 0 |

## Run Artifacts

| Artifact | Value |
|----------|-------|
| Conversation turns | 54 |
| Tool call traces | 49 |
| Generated files | 2 |
| Exit code | 0 |
| Completed at | 2026-10-05T04:04:43.549013+00:00 |

### Generated Files

- /home/daytona/app/requirements.txt
- /home/daytona/app/generate_image.py

## Related Pages

- [Codex API leaderboard](/leaderboard/codex/api)
- [Compare Codex API companies](/leaderboard/codex/api/compare)
- [Agent Landscape](/leaderboard/discoverability)

Canonical URL: https://devtoolarena.com/codex/api/replicate