text model
Gemini 3.5
GoogleHuge context and native audio/video understanding at aggressive prices.
Gemini 3.5 is the value pick when the input is large or non-textual. Its native audio and video understanding is why the same family powers our transcription tier — you can hand it a raw interview recording rather than a transcript.
Variants
3 ways to run it.
Same family, different quality/cost rung. Picking the wrong one is the most common way teams overpay.
Pro
Highgemini-3.5-proFrontier tier with the full 1M context.
Flash
Balancedgemini-3.5-flashThe price/performance sweet spot.
Flash-Lite
Minigemini-3.5-flash-liteCheapest tier for mechanical work.
Fit
Two audiences, one model.
The same model is a different proposition depending on what you point it at.
Recruitment fit
The cheapest credible way to process raw interview recordings without a separate transcription step.
Best for
- Interview recording analysis
- Multilingual pipelines
- Bulk CV parsing
Strengths
- Accepts audio and video directly — no transcribe-then-reason hop
- 1M context swallows an entire hiring round at once
- Best measured Greek accuracy of anything we have benchmarked
Watch out for
- Reasoning trails Opus 5 and GPT-6 on rubric consistency
The numbers
Measured, not asserted.
HumanLike evals
Run on our own harness.
- Greek transcription WER (Live sibling)
- 0.03Our eval
Measured on gemini-3.5-transcribe-live, the audio sibling of this family.
