Meta models
8models tracked · ranked by intelligence, speed & price
Match · Updated July 2026
Meta's smartest model is Llama 4 Maverick (14.3 on the Intelligence Index), and its cheapest is Llama 3.2 1B Instruct at $0.027 per million input tokens. All 8 Meta models are compared below by intelligence, speed, latency, context and price.
Smartest MetaFastest MetaCheapest MetaSmall & fast MetaMeta for codingMeta reasoningMeta for designMeta for agents
Smartest
14.3
meta-llama/llama-4-maverick
Fastest
—
Cheapest
$0.027/M
meta-llama/llama-3.2-1b-instruct
| Model | Intel | Speed | Latency | In $/M | Context |
|---|---|---|---|---|---|
| llama-4-maverick ToolsJSONVision | 14.3 | — | — | $0.200 | 1.0M |
| llama-4-scout ToolsJSONVision | 10.0 | — | — | $0.100 | 1.3M |
| llama-guard-4-12b JSONVision | — | — | — | $0.180 | 1.0M |
| llama-3.3-70b-instruct ToolsJSON | — | — | — | $0.130 | 131K |
| llama-3.2-1b-instruct | — | — | — | $0.027 | 60K |
| llama-3.2-3b-instruct JSON | — | — | — | $0.050 | 131K |
| llama-3.1-70b-instruct ToolsJSON | — | — | — | $0.400 | 131K |
| llama-3.1-8b-instruct ToolsJSON | — | — | — | $0.050 | 131K |