GPT-5 Codex API: Endpoint & Pricing
Deprecated 23 Jul 2026GPT-5 Codex is not a publicly released or documented model from OpenAI, and no reliable technical or capability information is available about it.
What is GPT-5 Codex?
GPT-5 Codex is an unreleased and undocumented model name attributed to OpenAI for which no official information currently exists. Because of this, there are no confirmed details about its intended use cases or capabilities. There are likewise no authoritative statements about its relationship to prior OpenAI model families such as GPT or Codex.
Can you self-host GPT-5 Codex?
No. We found no public weights for GPT-5 Codex on Hugging Face, so there is nothing to download, quantize or serve on your own GPUs: no VRAM budget, no vLLM or GGUF build to plan for. The only way to run it is over an API.
Skip the deploy — use LLM.API as your endpoint
No GPUs, no quantization trade-offs. Same model, OpenAI-compatible, with custom volume discounts on request.
Providers
Different companies host the same model. OpenRouter routes your request to one of them based on the routing mode you pick — Balanced (price + speed), Nitro (fastest), or Exacto (highest tool-calling accuracy).
List price by provider
InputOutputProvider list prices; the LLM.API discount applies on top.
| Provider | Input | Output | Cache read /M | Latency | Throughput | Uptime |
|---|---|---|---|---|---|---|
| in $1.25; out $10 per 1M tokens | 400K tokens | vision, tools, streaming, reasoning, JSON, structured |
Prices and availability from the LLMAPI catalogue, updated nightly. Last updated 11 Oct 2026.
Try this model
Test GPT-5 Codex right here — free to start.
Suggestions for your first prompt
Code snippet
Call the model through the OpenAI-compatible API.
Switching from OpenAI? Change 2 lines.
- base_url="https://api.openai.com/v1"
- api_key="YOUR_OPENAI_KEY"
+ base_url="https://api.llmapi.ai/v1"
+ api_key="YOUR_LLMAPI_KEY"Everything else stays the same — SDK, message format, tools, streaming. Set model to gpt-5-codex and the call runs unchanged.
curl https://api.llmapi.ai/v1/chat/completions \
-H "Authorization: Bearer $LLMAPI_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-5-codex",
"messages": [
{"role": "user", "content": "Give me three crisp launch checklist items."}
],
"reasoning_effort": "low"
}'from openai import OpenAI
client = OpenAI(
api_key="YOUR_LLMAPI_KEY",
base_url="https://api.llmapi.ai/v1",
)
resp = client.chat.completions.create(
model="gpt-5-codex",
messages=[
{"role": "system", "content": "You are a precise product assistant."},
{"role": "user", "content": "Give me three crisp launch checklist items."},
],
reasoning_effort="low",
)
print(resp.choices[0].message.content)import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.LLMAPI_KEY,
baseURL: "https://api.llmapi.ai/v1",
});
const resp = await client.chat.completions.create({
model: "gpt-5-codex",
messages: [
{ role: "user", content: "Give me three crisp launch checklist items." },
],
});
console.log(resp.choices[0].message.content);stream = client.chat.completions.create(
model="gpt-5-codex",
messages=[{"role": "user", "content": "Write a haiku about latency."}],
stream=True,
)
for chunk in stream:
delta = chunk.choices[0].delta.content
if delta:
print(delta, end="", flush=True){
"model": "gpt-5-codex",
"messages": [
{"role": "system", "content": "You are a precise product assistant."},
{"role": "user", "content": "Give me three crisp launch checklist items."}
],
"reasoning_effort": "low"
}ENDPOINTS
Supported endpoints & features
Base URL https://api.llmapi.ai/v1 · model ID gpt-5-codex · OpenAI-compatible.
POST /v1/chat/completionsChat Completions — supported
| Feature | Status | How to set it |
|---|---|---|
| Function calling (tools) | Supported | tools, tool_choice |
| Parallel tool calls | Not supported | parallel_tool_calls |
| JSON mode | Supported | response_format: {"type": "json_object"} |
| Structured outputs | Supported | response_format: {"type": "json_schema", …} |
| Streaming | Supported | stream: true |
| Vision (image input) | Supported | image_url content parts |
| Web search | Not supported | web_search |
| Reasoning effort | low · medium · high | reasoning_effort: "low" |
ERRORS
Errors & fallback routing
| Code | What it means | What to do |
|---|---|---|
401 | Missing or invalid API key. | Send Authorization: Bearer YOUR_LLMAPI_KEY and check the key is active. |
429 | Rate limited, or the upstream provider is throttling the request. | Back off and retry with jitter; the gateway also retries the request on another provider where one is available. |
5xx | Upstream provider error or timeout. | Retry; fallback routing sends the retry to the next healthy provider for this model. |
When a provider fails or throttles, the request is routed to the next provider serving this model — the ones listed in the providers table above.
Uptime
Last 30 days
26/30 days operational | 96.70% uptime
GPT-5 Codex in production
What the model is good at.
-
Conversational AI
Engages in multi-turn dialogue, answering questions and following instructions across many topics in clear, coherent natural language.
-
Language Translation
Translates text between multiple languages while preserving meaning, tone, and essential formatting for general-purpose use cases.
-
Text Analysis
Analyzes user-provided text to extract key points, summarize content, and support tasks like classification or information organization.
-
Code Reasoning
Understands and explains source code, assisting with debugging, refactoring ideas, and conceptual clarification based on textual descriptions.
-
Image Reasoning
Interprets user-supplied images to support tasks like description, object identification, and contextual reasoning, when such inputs are available.
What you get when you call GPT-5 Codex through LLM.API.
-
Unified AI Routing
Dynamically route each request to the optimal model across providers based on latency, cost, and quality—without changing your code or deployment pipeline.
One API, any model -
Cost-Aware Orchestration
Automatically balance performance and price using configurable policies, so you avoid overpaying for premium models while keeping SLAs and quality intact.
Optimize every token -
Resilient Fallback Flows
Survive provider outages and rate limits with automatic failover to backup models, preserving uptime and user experience without manual incident playbooks.
Never ship a dead endpoint -
End-to-End Observability
Track latency, cost, and model behavior in one place with request-level traces, logs, and metrics that plug cleanly into your existing monitoring stack.
See every token’s path -
Task-Level Abstractions
Define tasks like chat, RAG, or classification once, then swap models or providers freely while keeping consistent inputs, outputs, and evals.
Program tasks, not models -
High-Throughput Batch
Run massive offline jobs with automatic chunking, retries, and concurrency control, achieving cloud-scale throughput without writing custom batch infrastructure.
Batch at cloud scale
When to Use — When NOT to Use
Use it if...
- You need a strong general-purpose model from OpenAI for versatile coding assistance.
- You need tight integration with the broader GPT-5 ecosystem and OpenAI tooling.
- Your use case involves prototyping AI-powered developer tools that leverage advanced language understanding.
- You need reliable code completion, explanation, and refactoring across multiple popular programming languages.
- Your use case involves combining natural language reasoning with code generation in the same workflow.
- You need a single model that can handle code plus general text tasks effectively.
Avoid if...
- You need strict on-prem or air-gapped deployment where cloud-hosted OpenAI models are disallowed.
- You need a highly specialized model fine-tuned on proprietary domain data only you control.
- Your workload requires the absolute lowest possible latency from an on-device or edge model.
- You need deterministic, fully reproducible outputs for safety-critical code generation without human review.
- Your workload requires avoiding reliance on any third-party hosted AI provider for compliance reasons.
- You need a tiny, resource-constrained model that can run efficiently on microcontrollers.
COMMUNITY
What developers say about GPT-5 Codex
— No public developer feedback about this model yet.
Frequently Asked Questions
-
What is GPT-5 Codex?
GPT-5 Codex is an OpenAI code-focused large language model, optimized for program synthesis, refactoring, and natural-language-to-code workflows via LLM.API.
-
What is GPT-5 Codex best at?
GPT-5 Codex excels at generating production-grade code, explaining complex codebases, automated refactoring, and creating end-to-end implementations from natural language specifications.
-
How is GPT-5 Codex priced on LLM.API?
GPT-5 Codex pricing on LLM.API is usage-based per token, with exact input and output rates defined in your LLM.API pricing dashboard.
-
What is the context window of GPT-5 Codex?
GPT-5 Codex supports a large context window suitable for multi-file repositories; check the LLM.API model card for the current maximum token limit.
-
How fast is GPT-5 Codex in terms of latency?
GPT-5 Codex typically returns initial tokens within a few seconds, with total latency depending on prompt size, response length, and current LLM.API load.
-
Which modalities does GPT-5 Codex support?
GPT-5 Codex supports text prompts and text outputs, and is optimized specifically for source code and natural-language instructions.
-
How do I access GPT-5 Codex through LLM.API?
You call the LLM.API chat or completion endpoint with the GPT-5 Codex model identifier, using your LLM.API key for authentication.
-
How does GPT-5 Codex compare to general-purpose GPT-5 models?
Compared to general-purpose GPT-5 variants, GPT-5 Codex is more capable and reliable on code tasks but less optimized for open-ended conversational content.
-
What limitations does GPT-5 Codex have?
GPT-5 Codex can still produce incorrect or insecure code, may hallucinate APIs, and does not automatically validate, test, or run generated programs.
-
Can GPT-5 Codex work with entire repositories or large codebases?
GPT-5 Codex can handle large code snippets and summaries of repositories within its context window, but full monorepos may require chunking and tooling integration.
COMPARE
Competitive Models
GPT-5
$1.25 in / $10 out per 1M tokens, 400,000 context — the closest current alternative to GPT-5 Codex.
Qwen3 Max 2026-01-23
$1.2 in / $6 out per 1M tokens, 262,144 context — the closest current alternative to GPT-5 Codex.
GLM-5-Turbo
$1.2 in / $4 out per 1M tokens, 200,000 context — the closest current alternative to GPT-5 Codex.
text-embedding-ada-002 is an OpenAI embedding model that converts text into numerical vectors for measuring semantic similarity. It is an improved, more performant successor to earlier Ada-based embedding models and became a widely used default for production applications.
Whisper 1
Whisper 1 is OpenAI’s hosted automatic speech recognition model based on the open-source Whisper family, designed for high-quality transcription and translation of audio. It is notable for robust multilingual speech-to-text performance and language identification across diverse audio conditions.
Text Embedding 3 Large
Text Embedding 3 Large is OpenAI’s high‑capacity embedding model optimized for semantic search, retrieval, and clustering tasks. It provides high‑quality vector representations of text with strong performance across diverse domains.
Get one key to every model
Swap your API key. Keep your code.