Not every project needs flagship pricing. Gemini 3.5 Flash and GPT-5.6 Luna are the top budget contenders.
Gemini 3.5 Flash
BenchLM 75.5. Fastest inference of any model. Native multimodal. Flash-Lite variant for extreme cost savings.
GPT-5.6 Luna
$1/$6 per million tokens. Same architecture as Sol, scaled down. Strong function calling and structured outputs.
| Dimension | Flash | Luna |
|---|---|---|
| Input Price | $0.15/M | $1/M |
| Output Price | $0.60/M | $6/M |
| BenchLM | 75.5 | ~55 |
Winner
Flash for speed, multimodal, and value. Luna for OpenAI ecosystem compatibility.
Related: Best Free AI Models · Free AI Tools
