AI Model API IDs: Developer Reference (2026)

The current flagship API IDs are gpt-5.6-sol, claude-fable-5, gemini-3.5-flash, and grok-4.5. This developer reference lists exact model strings for current text, coding, agent, and image generation APIs. Every ID below was checked against first-party provider documentation on July 18, 2026.

Model display names are not always valid API values. GPT-5.6 is an alias for GPT-5.6 Sol, and Google’s Nano Banana names map to Gemini image model IDs. Use the strings in the API model ID column in your code.

AI Model API ID Quick Reference

These are the current developer-facing model IDs worth starting with. Choose a flagship model for difficult work, a balanced model for most production traffic, and a small model for routing or repetitive tasks.

OpenAI Model IDs

OpenAI’s current GPT-5.6 family has three production tiers. The official catalog recommends Sol for complex reasoning and coding, Terra for balance, and Luna for cost-sensitive volume.

• gpt-5.6-sol: flagship model. The shorter gpt-5.6 alias currently points to Sol.
• gpt-5.6-terra: balanced intelligence and cost.
• gpt-5.6-luna: lowest-cost GPT-5.6 tier for high volume.

All three support text and image input, text output, reasoning controls, function calling, web search, file search, and computer use through the Responses API. See the official OpenAI model catalog for the live capability matrix.

Anthropic Claude Model IDs

Anthropic’s current production lineup is Fable 5, Opus 4.8, Sonnet 5, and Haiku 4.5. Fable 5 is the highest-capability option, while Opus 4.8 is Anthropic’s recommended starting point for complex agentic coding and enterprise work.

• claude-fable-5: highest-capability long-running agents.
• claude-opus-4-8: complex coding and enterprise work.
• claude-sonnet-5: best general speed and intelligence mix.
• claude-haiku-4-5-20251001: pinned fast model. The moving alias is claude-haiku-4-5.

Anthropic also lists claude-mythos-5, but it is limited to approved Project Glasswing customers rather than general API access. The full availability table is in the official Claude models overview. If you are implementing Claude in a coding workflow, my complete Claude Code guide covers the surrounding tools and configuration.

Google Gemini Model IDs

Gemini 3.5 Flash is Google’s stable frontier model for sustained agentic and coding performance. Gemini 3.1 Pro remains a preview model, while Gemini 3.1 Flash-Lite is the stable efficiency option.

• gemini-3.5-flash: stable frontier text model.
• gemini-3.1-pro-preview: preview model for complex reasoning.
• gemini-3.1-flash-lite: stable high-volume model.

Google recommends stable IDs for production. A -latestalias can be hot-swapped, and preview models typically have shorter deprecation windows. Check the Gemini API models page before a production launch.

xAI Grok Model IDs

Grok 4.5 uses the API ID grok-4.5. xAI positions it as the current frontier model for coding, agentic tasks, and knowledge work, with both Responses API and Chat Completions support.

xAI’s API is OpenAI SDK-compatible when you set the base URL to https://api.x.ai/v1. The official Grok 4.5 developer guide includes SDK and curl examples.

Image Generation Model IDs

Nano Banana 2 uses gemini-3.1-flash-image, Nano Banana 2 Lite uses gemini-3.1-flash-lite-image, and Nano Banana Pro is gemini-3-pro-image. Google recommends the Nano Banana family for most image generation and editing workflows.

OpenAI also publishes the pinned GPT Image 2 snapshot gpt-image-2-2026-04-21. Use the moving gpt-image-2 alias when you want automatic upgrades, or the dated snapshot when visual consistency matters. For a real image API implementation, see my image generation MCP project and the behind-the-scenes Dream Pixel Forge build notes.

Copy-Paste API Examples

Put the exact model ID in the provider SDK’s model field. These minimal JavaScript examples use each provider’s current primary API.

OpenAI Responses API

import OpenAI from "openai";

const client = new OpenAI();
const response = await client.responses.create({
  model: "gpt-5.6-sol",
  input: "Design a resilient job queue.",
});

console.log(response.output_text);

Anthropic Messages API

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic();
const response = await client.messages.create({
  model: "claude-opus-4-8",
  max_tokens: 2048,
  messages: [{ role: "user", content: "Review this architecture." }],
});

Google Gemini API

import { GoogleGenAI } from "@google/genai";

const client = new GoogleGenAI({});
const response = await client.models.generateContent({
  model: "gemini-3.5-flash",
  contents: "Plan a migration with rollback steps.",
});

console.log(response.text);

Google Nano Banana 2

const interaction = await client.interactions.create({
  model: "gemini-3.1-flash-image",
  input: "Create a clean technical diagram of a queue worker.",
  response_format: { type: "image", aspect_ratio: "16:9" },
});

Aliases vs Pinned Snapshots

Use a stable alias when you want provider improvements without changing your code. Use a dated snapshot when repeatability matters more than automatic upgrades, especially for evaluated prompts, regulated workflows, and image generation with strict visual acceptance tests.

• Moving alias example: gpt-5.6 currently resolves to GPT-5.6 Sol.
• Stable model example: gemini-3.5-flash is Google’s named stable release.
• Pinned snapshot example: gpt-image-2-2026-04-21 locks GPT Image 2 behavior to a dated version.
• Preview example: gemini-3.1-pro-preview can change or deprecate faster than a stable model.

Before switching a production model, run the same eval set against both IDs. Display-name upgrades can change output quality, tool behavior, latency, and cost even when the request schema stays the same. My Claude Code vs Codex vs Cursor comparison shows how model differences surface in real coding workflows.

Discover Available Models

Provider model-list endpoints are the final authority for your account. Availability can vary by region, organization, access tier, and cloud platform, so query the API when an otherwise valid ID returns a not-found error.

# OpenAI
curl https://api.openai.com/v1/models \
  -H "Authorization: Bearer $OPENAI_API_KEY"

# Google Gemini
curl "https://generativelanguage.googleapis.com/v1beta/models?key=$GEMINI_API_KEY"

# xAI
curl https://api.x.ai/v1/models \
  -H "Authorization: Bearer $XAI_API_KEY"

Official Sources

This page uses only first-party model documentation. Recheck these catalogs before committing an ID to a long-lived integration.

• OpenAI API model catalog
• Anthropic Claude models overview
• Google Gemini API models
• Google Nano Banana image generation guide
• xAI model catalog