Google AI Models
8 models from Google on the AMR board, ranked by AMR score. Snapshot 2026-11-06.
Best on the board: Gemini 4 Argon at #7 — 81.0 AMR
| Rank | Model | AMR Score | GPQA | Context | Price (in / out) | Speed | Released |
|---|---|---|---|---|---|---|---|
1▲2 | Gemini 4 Argon | 81.0 | 92.1% | 1.0M | $2 / $10 | 165 | Sep 30 |
2— | Gemini 4.0 | 76.4 | 87.9% | 1.0M | $2.20 / $10 | 158 | Jul 8 |
3NEW | Gemini 4 Carbon | 74.4 * | — | 1.0M | $10 / $50 | — | Oct 28 |
4▲2 | Gemini 3.8 Flash | 69.3 | 78.9% | 1.0M | $0.75 / $3.75 | 342 | Sep 18 |
5— | Gemini 3 Pro | 69.0 | 80.9% | 1M | $1.50 / $6 | 168 | Feb 11 |
6— | Gemini 3.1 Pro | 68.3 | 79.8% | 1.0M | $2 / $12 | 172 | Apr 21 |
7— | Gemma 4 27B | 60.7 | 67.4% | 262K | $0.09 / $0.30 | 174 | Jul 24 |
8NEW | Pixel Canary | 38.6 | 31.2% | 32K | Free / Free | 210 | Sep 26 |
* Provisional score — benchmarks pending
Google pricing sources
The official pricing page lists per-model rates. Check it there before trusting any number here.
Paid tier rates per 1M tokens by model and context size; free tier has separate quotas.
Pricing page last checked 2026-10-28
Coding plan · Gemini Code Assist
Coding assistant across the Gemini subscription tiers, with IDE and terminal agents.
See Gemini Code Assist pricing ↗Share the Google lineup
Copy the card below — or grab the link — straight into Slack, X, Discord or Reddit.