Gemini 3.8 Live Extended Thinking

GA

GA/stable Live API model: Google lists it under "Gemini 3 / Stable" on the models index with the badge "New - Stable", and the model page's Versions row reads "Stable: gemini-3.8-live-extended-thinking". Google describes it as the "High-reasoning Live API model for voice interactions, recommended when higher background reasoning is required", and says it "processes background reasoning and asynchronous tool calls while streaming continuous audio responses". ONE SHARED PRICING TABLE COVERS THREE MODEL IDS: the pricing page heads a single Standard table with "Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, and Gemini 3.1 Flash Live Preview", so these rates are not stated for this model alone. Rates differ per input type: input $0.75 (text) / $3.00 or $0.005 per minute (audio) / $1.00 or $0.002 per minute (image/video); output $4.50 (text) / $12.00 or $0.018 per minute (audio). Values shown are the TEXT tier, matching how the catalog carries the other multi-tier Gemini rows. No context-caching rate is published and the spec table states caching is "Not supported", so the cached-input field is null as a sourced absence rather than as an unknown. Free tier is "Free of charge". Grounding with Google Search: 5,000 free requests per month shared across Gemini 3.x, then $14 per 1,000 requests. PRICED IDENTICALLY TO gemini-3.8-live ON EVERY AXIS - the extra reasoning carries no rate premium in what Google publishes, because the two models share that one table. Spec table gives 131,072 input / 65,536 output tokens, inputs text/images/audio/video and outputs text and audio; audio generation, thinking, search grounding and the Live API are supported, function calling is supported ASYNC ONLY (the one capability line that differs from gemini-3.8-live, which also accepts blocking calls), and caching, code execution, file search, Maps grounding, image generation, structured outputs, URL context and the Batch API are not. Google does NOT name this model as a migration target for any endpoint; the four Live endpoints it re-pointed on 2026-09-15 all point at gemini-3.8-live. Released 2026-09-15 per the Release date column of Google's deprecations table, which states "No shutdown date announced" for it; the row is NOT gray, so nothing here is inferred from an absence. Google publishes no knowledge cutoff on any surface, so that stays null. All three Google surfaces read for this row - the models index, the model page and the pricing page - are stamped "Last updated 2026-09-15 UTC". Lifecycle read first-party from Google's deprecations table on 2026-09-16.

Gemini 3.8 Live Extended Thinking by Google costs $0.75 per 1M input tokens and $4.50 per 1M output tokens ($1.69/1M blended), with a 131K (131.072-token) context window. It is generally available (GA).

Last verified: 26 Sep 2026 · sourced from official provider documentation

Provider
Google
Status
GA
Input price
$0.75 / 1M tokens
Output price
$4.50 / 1M tokens
Cached input
—
Blended price
$1.69 / 1M tokens
Context window
131.072 tokens (131K)
Max output
65.536 tokens
Modality
text, audio, video, image, audio-out
Open weights
No — API only
Knowledge cutoff
—
Released
15 Sep 2026
API string
gemini-3.8-live-extended-thinking

Source: Google official documentation ↗

FAQ

Gemini 3.8 Live Extended Thinking — questions & answers

How much does Gemini 3.8 Live Extended Thinking cost?

Gemini 3.8 Live Extended Thinking is priced at $0.75 per 1M input tokens and $4.50 per 1M output tokens ($1.69/1M blended at a 3:1 input-to-output ratio).

What is the context window of Gemini 3.8 Live Extended Thinking?

Gemini 3.8 Live Extended Thinking has a 131.072-token context window (131K), with up to 65.536 output tokens per request.

Is Gemini 3.8 Live Extended Thinking deprecated?

No — Gemini 3.8 Live Extended Thinking is generally available (GA) and not currently scheduled for deprecation or retirement in our tracker.

Does Gemini 3.8 Live Extended Thinking support image or vision input?

Yes — Gemini 3.8 Live Extended Thinking accepts image input. Listed modalities: text, audio, video, image, audio-out.

Can I self-host Gemini 3.8 Live Extended Thinking?

No — Gemini 3.8 Live Extended Thinking is proprietary. Its weights are not publicly released, so it is available only through Google's API (or hosted partners), not for self-hosting.

What is the API model string for Gemini 3.8 Live Extended Thinking?

The API model identifier for Gemini 3.8 Live Extended Thinking is "gemini-3.8-live-extended-thinking" when calling Google's API.