[]AI code editor· 6 models rankedMulti-vendor

Best Models for Cursor

The AI code editor — a curated cross-vendor frontier lineup

Unlike Claude Code or Codex, Cursor is not locked to one vendor — it exposes a curated set of frontier coding models from Anthropic, OpenAI, Google, and xAI, plus an Auto mode that routes each request to a reliable model for you. That makes model choice a cross-vendor decision. Claude Opus 4.8 is the standout for hard agentic work (and the most token-efficient flagship), with GPT-5.5 and Gemini 3.1 Pro as strong alternatives and faster tiers for everyday edits. Per-model pricing in Cursor is metered in credits and listed on Cursor's pricing page; we link it rather than reproduce per-token rates here.

Best pick

Claude Opus 4.8Cloud API

Best for hard, long-horizon agentic editing in Cursor — and the most token-efficient flagship: ~35% fewer output tokens per task than Opus 4.7 at the same rate.

What Cursor needs

A frontier model with strong multi-file reasoning and agentic edits; Auto mode is the safe default, while picking a specific model lets you optimize for capability, speed, or token efficiency.

Supported models, ranked

S
Claude Opus 4.8Cloud APIAnthropic · Flagship$5 / $25 $/MTok · 200K

Best for hard, long-horizon agentic editing in Cursor — and the most token-efficient flagship: ~35% fewer output tokens per task than Opus 4.7 at the same rate.

↘ −35% output tokens vs 4.7

fast mode: $10 / $50 per MTok

S
GPT-5.5Cloud APIOpenAI · Flagship— / — $/MTok

OpenAI's newest frontier model; an excellent all-round alternative for complex coding and computer-use tasks.

A
Gemini 3.1 ProCloud APIGoogle · Flagship— / — $/MTok

Google's frontier coding model — very large context and strong multi-file reasoning, a top pick for big repos.

A
GPT-5.3-CodexCloud APIOpenAI · Balanced— / — $/MTok

Codex-specialized OpenAI checkpoint, available in Cursor for coding-focused sessions.

B
Grok Build 0.1Cloud APIxAI · Balanced— / — $/MTok

xAI's coding-focused model — a newer alternative voice in the cross-vendor lineup.

B
Gemini 3.5 FlashCloud APIGoogle · Fast— / — $/MTok

Fast, low-cost Gemini tier for quick completions and high-volume agent turns.

How to use it

In Cursor, open the model picker and choose a model or leave it on Auto (Cursor selects by reliability and current performance, billed at flat credit rates). Pro is $20/month with a credit pool; pick Opus 4.8 or GPT-5.5 for the hardest tasks and a Flash/mini tier for quick completions.

Cursor official site ↗

The weekly local-AI refresh

New open-weight models, real Apple Silicon benchmarks, and the one model worth running on your Mac this week. Free, one email a week, unsubscribe anytime.

Frequently Asked Questions

What is the best model to use in Cursor?+
For the hardest agentic tasks, Claude Opus 4.8 leads and is the most token-efficient flagship. GPT-5.5 and Gemini 3.1 Pro are strong cross-vendor alternatives, and faster tiers like Gemini 3.5 Flash suit quick edits. If unsure, Auto mode picks a reliable model per request.
Does Cursor only run Claude models?+
No. Cursor is multi-vendor — it offers frontier models from Anthropic (Claude), OpenAI (GPT-5), Google (Gemini), and xAI (Grok), so you can switch vendors per task or let Auto mode choose.
How does Cursor bill for these models?+
Cursor meters usage in credits on top of its subscription (Pro is $20/month with a credit pool, Pro+ adds more). Per-model and Auto-mode rates are listed on Cursor's pricing page, which we link in Sources.

Sources

Other AI Coding Tools

>_

Claude Code

Cloud

Anthropic's terminal coding agent — pointed at a local model

$_

OpenCode

Open-source terminal coding agent (Ollama / OpenAI-compatible)

🐾

OpenClaw

Self-hosted agentic assistant with first-class Ollama support

A>

Aider

Terminal AI pair-programmer with its own edit-format leaderboard

{}

Cline

Autonomous coding agent for VS Code (formerly Claude Dev)

#>

Roo Code

Autonomous VS Code agent with modes (a Cline fork)

->

Continue.dev

Open-source chat, edit & autocomplete for VS Code / JetBrains

~>

Goose

Block's open-source autonomous developer agent (MCP-driven)

<>

Open Claude Code

Open-source Claude Code CLI reimplementation, run on local models

5>

Codex

Cloud

OpenAI's coding agent — ranked across the GPT-5 model lineup

Explore More

ModelFit Wizard

Match a model to your exact Mac

Model Families

Browse every local model

Best Coding Models

Top local coders by hardware

Benchmarks

Compare model performance