low · medium · high
Only provider-documented reasoning controls are shown; missing controls remain unpublished rather than inferred.
MODEL REFERENCE / GOOGLE
Google Live voice endpoint with continuous audio, background thinking and asynchronous agent tool use.
PRIMARY-SOURCE VERIFIED
Google documents a 131,072-token context window. Input modalities: text + image + audio + video. Output: text + audio. Pricing status: Official paid API.
CAPABILITY CONTRACT
Only provider-documented reasoning controls are shown; missing controls remain unpublished rather than inferred.
Input modalities come from the cited official model documentation.
Output modality and output-limit claims remain separate so an undocumented token cap is not guessed.
PRICING STATUS
$0.75 input · $4.50 output per million TEXT tokens; audio/video priced separately.
SXF preserves the provider-published native billing basis. Only calculator-eligible generative token models enter token-cost arithmetic; specialist units and input-only embeddings remain visible without forced conversion.
CAVEATS
Official API ID gemini-3.8-live-extended-thinking is distinct from gemini-3.8-live.
Standard TEXT input/output $0.75/$4.50 per million text tokens; AUDIO input/output $3/$12 per million audio tokens; image/video input $1 per million tokens.
Audio sessions use multiple native meters; text calculator disabled.
Only non-blocking async function calls supported. Clients must use interaction_status to detect a truly idle session.
EXPLORE
OFFICIAL SOURCES
WHAT CHANGED
No post-baseline factual changes recorded for Gemini 3.8 Live Extended Thinking since 2026-10-12.
VERIFIED CHANGE HISTORY
Baseline verified: 131,072 context · 65,536 max output · $0.75 input / $4.50 output per million TEXT tokens; audio/video priced separately
Official evidence ↗Append-only SXF ledger · 1 verified event · chain head 77b0ad167e15…
COMPARE GEMINI 3.8 LIVE EXTENDED THINKING