Best Model for Cursor

Claude Sonnet 5.5 is the best everyday model in Cursor: it scores 61.8% on Terminal-Bench 4.0, about three points behind Claude Opus 5.5, at half the price ($2/$10 per million tokens against $4/$20). Since October 12, 2026, Cursor's new plans include usage for its own Grok and Composer models only and bill Claude, GPT and Gemini at API list price on Premium ($30) and above, so Grok 4.7 (37.6% on Terminal-Bench 4.0) is the pick when you want to stay within the plan. Cursor's model list does not include the GPT-6 models; its newest OpenAI models are the GPT-5.6 tiers.

Read the full Cursor review → · updated 2026-10-17

Sources: Cursor models and pricing · Our Cursor pricing guide · Terminal-Bench 4.0 leaderboard · Anthropic on Haiku 5.5

Run Claude Sonnet 5.5, Claude Opus 5.5 and Claude Haiku 5.5 in your agent

Or let the router send each step to the cheapest model that can do it, on providers we vet for price, uptime and speed. Plans from $19 a month.

Cursor setup →See plans

#1Claude Sonnet 5.5Anthropic · $4.40/1M · 1M ctx

Best overall. 61.8% on Terminal-Bench 4.0 and 95.8 on Scale's SWE-Bench Pro V2 at $2/$10 per million tokens: the best balance of quality and cost for agent and chat work in Cursor.

Overall 96/100 · Value 59Full breakdown →

#2Claude Opus 5.5Anthropic · $8.80/1M · 1M ctx

Best for long agent runs. #1 on Terminal-Bench 4.0 at 64.9%. Switch to it for hard, multi-file changes; at $4/$20 it bills twice Sonnet 5.5 per token.

Overall 98/100 · Value 51Full breakdown →

#3Grok 4.7xAI · $3.20/1M · 500K ctx

Budget pick. Covered by the usage included in Cursor's plans, and the $8 Plus plan runs only Grok models. It scores 37.6% on Terminal-Bench 4.0, well behind the Claude models.

Overall 82/100 · Value 58Full breakdown →

#4Claude Haiku 5.5Anthropic · $0.22/1M · 1M ctx

Best value. Cursor lists it at $0.10/$0.50 per million tokens, a twentieth of Sonnet 5.5. Anthropic reports 39.2% on Terminal-Bench 4.0, enough for quick edits and questions.

Overall 82/100 · Value 97Full breakdown →

#5GLM 5.3Z.AI · $1.50/1M · 1.0M ctx

Open-weight pick. The top open-weight model on Terminal-Bench 4.0 (41.8%), listed in Cursor at $1.40/$4.40 per million tokens.

Overall 83/100 · Value 69Full breakdown →

FAQ

Which model is best in Cursor?

Claude Sonnet 5.5 for everyday agent work (61.8% on Terminal-Bench 4.0), Claude Opus 5.5 for the hardest tasks (64.9%), and Grok 4.7 when you want to stay within your plan's included usage.

Does the model choice affect what Cursor costs?

Yes. On the plans launched October 12, 2026, included usage covers Cursor's Grok and Composer models, and Claude, GPT and Gemini bill at API list price on top. Cursor's docs add a Cursor Token Rate on third-party models. Opus 5.5 costs twice Sonnet 5.5 per token, and Haiku 5.5 a twentieth.

Can I use my own API key in Cursor?

On Premium and above. Cursor Settings, Models, takes OpenAI, Anthropic, Google, Azure OpenAI and AWS Bedrock keys for chat and agent requests; Tab completions still run on Cursor's own models. The Override OpenAI Base URL setting also catches Grok and Composer, so turn it off to use your included Grok usage.

Can I use GPT-6 in Cursor?

Not as of October 13, 2026. Cursor's model list tops out at the GPT-5.6 tiers (Luna, Terra, Sol). For GPT-6.1 Sol in an editor, GitHub Copilot and Windsurf list it, and Cline or Kilo Code can call it with your own key.

Current Claude, GPT, Gemini and open models run on Standard Compute

Standard Compute

Current frontier models Claude Fable 5.1 · GPT-6 Astra