Standard Compute
← All models

Best Model for Gemini CLI

Gemini 3.8 Flash is the best model for most Gemini CLI work: it scores 94.9 on Scale's SWE-Bench Pro V2 and 41 on the Artificial Analysis Intelligence Index, ahead of Gemini 3.1 Pro Preview at 30, for an introductory $0.75/$3.75 per million tokens through December 31, 2026. Gemini CLI is built for Google's own models; Gemini 3.1 Pro Preview is the Pro option to try when Flash falls short on a hard, multi-step task. For long autonomous terminal runs, Gemini models trail the field: Gemini 3.8 Flash scores 19.1% on Terminal-Bench 4.0, where Claude Opus 5.5 in Claude Code scores 64.9%.

Read the full Gemini CLI review → · updated 2026-10-17

Sources: Gemini CLI model docs · Gemini API pricing · Scale SWE-Bench Pro V2 · Terminal-Bench 4.0 leaderboard · Source: Artificial Analysis (artificialanalysis.ai)

#1Gemini 3.8 FlashGoogle · $1.65/1M · 1.0M ctx

Best overall. Google's newest model, released September 2, 2026: 94.9 on Scale's SWE-Bench Pro V2 with a 1M-token context and multimodal input. The price doubles to $1.50/$7.50 on January 1, 2027.

Overall 76/100 · Value 63Full breakdown →
#2Gemini 3.1 ProGoogle · $5.00/1M · 1.0M ctx

Pro option. Gemini CLI's docs suggest switching to Pro when Auto or Flash isn't giving the results you want. It is still a preview model, and prompts over 200K tokens bill at $4/$18 per million.

Overall 64/100 · Value 44 · 29% community winFull breakdown →
#3Gemini 3.1 Flash LiteGoogle · $0.63/1M · 1.0M ctx

Budget pick. The cheapest Gemini with a 1M-token context, for quick, simple steps. Gemini CLI maps its Flash Lite option to Gemini 3.5 Flash Lite on accounts that have it.

Overall 55/100 · Value 58Full breakdown →

FAQ

Which Gemini model should I use in Gemini CLI?
Gemini 3.8 Flash for everyday work: it scores 94.9 on Scale's SWE-Bench Pro V2 and costs $0.75/$3.75 per million tokens until December 31, 2026. Switch to Gemini 3.1 Pro Preview for a hard task Flash gets wrong, and to Flash Lite for quick, simple steps. Gemini CLI's docs recommend the Auto setting, which picks between models per task.
Is Gemini CLI free?
Not any more. Since June 18, 2026 Google no longer serves Gemini CLI to free accounts or to Google AI Pro and Ultra subscribers; it moved them to Antigravity CLI, which has a free plan with weekly limits. Gemini CLI now runs on a Gemini Code Assist Standard or Enterprise licence or a paid Gemini API key, billed at Gemini API rates for the model you use. Its README still shows the old 1,000-requests-a-day free tier.
Can Gemini CLI use Claude or GPT models?
No. Gemini CLI is built for Google's Gemini models and only speaks the Gemini API. If you want Claude, GPT or open-weight models in a terminal agent, Qwen Code (which began as a Gemini CLI fork), OpenCode, Aider and Cline let you choose the provider.
Is Gemini 3.5 Flash still the default?
Gemini CLI's source now maps its Flash option to Gemini 3.8 Flash on accounts that have access, replacing Gemini 3.5 Flash. Run /model to see which models your account offers.

The agent matters as much as the model. Compare AI agents →

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

Standard Compute

Current frontier models Claude Fable 5.1 · GPT-6 Astra

More intelligence
per dollar.

Each step goes to the cheapest model that can do it, on providers we vet for price, uptime and speed, with automatic fallback when one fails.

Compliance
Opt-in rules keep a company, team or key on European or US providers
Privacy
We don’t train on your prompts or store them

Fixed monthly plans from $19 · cancel anytime

7-day fair refund Free trial, no card Billing by Stripe4,100+ agent users