GoogleAudio modelActive
Gemini 3.8 Live Extended Thinking
PopularityRank 670 of 875
0out of 100same as last month
Summary
Gemini 3.8 Live Extended Thinking targets complex, multi-step voice interactions that need background reasoning. Google describes it as an audio-to-audio model that reasons and runs asynchronous tool calls while streaming continuous audio. It accepts text, images, audio, and video and returns text and audio. The conversation can stay live while the model works through a harder request.
Reasoning effort can be set to low, medium, or high, and the API documents 131,072 input tokens with 65,536 output tokens. Compared with Gemini 3.8 Live, this variant keeps asynchronous function calls and their reasoning signals available while the audio response continues.
All benchmark results
Every published result with its source, test setup and comparison models.
Show all 47 published benchmark results
Page 1 of 4
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
Artificial Analysis Big Bench Audio speech reasoning (Show measurement, test conditions, and source)
- Source value
- 98
- Score
- 98
- Metric
- accuracy
- Unit
- %
- Benchmark version
- Big Bench Audio v1.2
- Category
- audio
- Direction
- Higher is better
- Sample
- 1000
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Independent evaluation. Native audio-to-audio on Big Bench Audio v1.2, 1,000 questions over four 250-item reasoning categories. Exact configuration: Gemini 3.8 Live Extended Thinking (High).
Artificial Analysis Conversational Dynamics (Show measurement, test conditions, and source)
- Source value
- 91.9
- Score
- 91.9
- Metric
- conversational behavior score
- Unit
- %
- Benchmark version
- AA Full Duplex Bench v1 + v1.5 view
- Category
- audio
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Independent evaluation. Full Duplex Bench v1/v1.5 conversational behavior view (pause handling, turn taking, interruption and backchannel categories). Exact configuration: Gemini 3.8 Live Extended Thinking (High).
Artificial Analysis τ-Voice task completion (Show measurement, test conditions, and source)
- Source value
- 68.6
- Score
- 68.6
- Metric
- Task Completion pass@1
- Unit
- %
- Benchmark version
- Artificial Analysis τ-Voice base task set
- Category
- audio
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Independent evaluation. Customer support agent task completion across airline (50), retail (114), telecom (114) scenarios, with tools and policy. Publisher's general protocol averages three trials; exact row Gemini 3.8 Live Extended Thinking (High) uses 3 trials.
Speech Agent Arena task success rate (Show measurement, test conditions, and source)
- Source value
- 89.1
- Score
- 89.1
- Metric
- eligible-conversation task success rate
- Unit
- %
- Benchmark version
- Current Speech Agent Arena board retrieved 2026-10-15
- Category
- audio
- Direction
- Higher is better
- Lower confidence bound
- 84.7
- Upper confidence bound
- 92.4
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Independent evaluation. Correct final task-completing tool call on eligible conversations only; participant deviations and unverifiable cases are excluded. Exact configuration: Gemini 3.8 Live Extended Thinking (High).
Sierra τ³-Banking task-completion score (Show measurement, test conditions, and source)
- Source value
- 35.1
- Score
- 35.1
- Metric
- reported banking-agent task completion
- Unit
- %
- Benchmark version
- Official Google DeepMind chart on Gemini Audio Live dialogue page, retrieved 2026-10-17
- Category
- agentic
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Google DeepMind Gemini Audio performance chart
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Vendor-reported result. Google DeepMind's official performance chart labels this result as the Sierra τ³-Banking leaderboard and names the exact model variant Gemini 3.8 Live Extended Thinking. The chart caption does not state reasoning effort, model/serving revision, task setup, sample count, or confidence interval. No Sierra evaluator row was located in this pass; this is retained solely as the official vendor-published report of its external benchmark result.
- Context
- This is Google DeepMind's vendor-reported chart value, not a direct Sierra leaderboard capture. Reasoning effort, evaluation setup, N, CI, and serving revision are not published in the page. Do not equate τ³-Banking with Artificial Analysis τ-Voice.
Cekura Gemini 3.8 Live Extended Thinking scenario reliability (pass³) (Show measurement, test conditions, and source)
- Source value
- 52.4
- Score
- 52.4
- Metric
- share of scenarios passing on all three runs (pass^3)
- Unit
- %
- Benchmark version
- Cekura Gemini Live benchmark, page dated 2026-10-16
- Category
- speech-to-speech
- Direction
- Higher is better
- Sample
- 82
- Source type
- independent-evaluator
- Evaluator
- Cekura speech-to-speech benchmark
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Cekura's separate row is Gemini 3.8 Live Extended Thinking, with thinking level high. The benchmark uses the same 82 scenarios for every model, each run three times, for 246 live calls per model. Reliability is the pass^3 scenario rate; the page reports 52.4%.
- Context
- The API serving revision is not pinned and the page does not publish a confidence interval. Do not confuse scenario N=82 with 246 calls.
Cekura TurnBench end-of-turn score (six-condition average) (Show measurement, test conditions, and source)
- Source value
- 84
- Score
- 84
- Metric
- end-of-turn score averaged across six line conditions
- Unit
- %
- Benchmark version
- Cekura TurnBench leaderboard, retrieved 2026-10-17
- Category
- speech-to-speech / turn-taking
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Cekura TurnBench
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. The exact leaderboard row is Gemini 3.8 Live Ext. Thinking. The source reports the end-of-turn score from Sesame TurnBench's published scorer, averaged over six line conditions and a 25-script private leaderboard set. It reports 36 calls per model per condition; all models heard the same recorded calls, streamed from US East with provider-default turn detection.
- Context
- The private benchmark set cannot be rerun externally. The page says its 95% range overlaps with several models but gives no exact interval for this row; no CI or serving revision is recorded.
MIVAS cross-industry pass@1 (Show measurement, test conditions, and source)
- Source value
- 81.9
- Score
- 81.9
- Metric
- cross-industry pass@1, task success on one scored trial
- Unit
- %
- Benchmark version
- MIVAS September 2026 evaluation; three-industry cross-industry row
- Category
- speech-to-speech / multi-agent task success
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Bluejay MIVAS
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. The exact MIVAS model profile is Gemini 3.8 Live Extended Thinking with model ID gemini-3.8-live-extended-thinking, HIGH thinking level and native audio input/output. The Bluejay LiveKit multi-agent voice-agent harness was evaluated in September 2026 on healthcare, legal and customer-support suites, 1,080 scored conversations in total with five scored trials per task. The source's direct Cross-industry row reports pass@1. No metric-specific denominator is published.
- Context
- This is a multi-agent system result, not an isolated model score. The public profile identifies the model and HIGH effort, but no immutable result-code revision or metric-specific N/CI is published.
MIVAS cross-industry pass^5 (Show measurement, test conditions, and source)
- Source value
- 55.5
- Score
- 55.5
- Metric
- cross-industry five-run reliability (pass^5)
- Unit
- %
- Benchmark version
- MIVAS September 2026 evaluation; three-industry cross-industry row
- Category
- speech-to-speech / multi-agent reliability
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Bluejay MIVAS
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. The exact MIVAS model profile is Gemini 3.8 Live Extended Thinking with model ID gemini-3.8-live-extended-thinking and HIGH thinking level. Bluejay reports three September 2026 industry suites, 1,080 scored conversations in total with five scored trials per task. The source's direct Cross-industry row reports pass^5; no metric-specific denominator is published.
- Context
- This is a multi-agent system result, not an isolated model score. The public profile identifies model and HIGH effort, but no immutable result-code revision or metric-specific N/CI is published.
Cekura agent-workflow success rate (Show measurement, test conditions, and source)
- Source value
- 76
- Score
- 76
- Metric
- calls passing every check
- Unit
- %
- Benchmark version
- Cekura speech-to-speech agent comparison, page last updated 2026-10-07
- Category
- speech-to-speech / agent workflow
- Direction
- Higher is better
- Sample
- 246
- Source type
- community-reported
- Evaluator
- Cekura speech-to-speech agent benchmark
- Status
- active
- Retrieved at
- 2026-10-18
- Methodology
- Community-reported result. Cekura speech-to-speech whole-agent workflow comparison. Exact model ID gemini-3.8-live-extended-thinking (displayed as Gemini 3.8 Live Extended Thinking); thinking level high; voice Charon; service turn detection; audio 16 kHz. Both compared systems use the same open-source Pipecat pipeline, prompt, tools, and connection. 82 scenarios run three times each; the page reports 2,952 calls across the benchmark.
- Context
- This is an agent-workflow result, not a model-only score. The page gives 82 scenarios and three runs per model; no 95% CI or immutable serving revision is published. The evaluator's institutional independence and affiliations are not established by this page. Data accuracy is conditioned on calls that save data, with the subset denominator unknown. Results apply only to the exact agent configuration above. N=246 calls per model from 82 scenarios x 3 runs; call-based measure.
Cekura agent replies reached expected outcome (Show measurement, test conditions, and source)
- Source value
- 88.6
- Score
- 88.6
- Metric
- agent-response outcome rate
- Unit
- %
- Benchmark version
- Cekura speech-to-speech agent comparison, page last updated 2026-10-07
- Category
- speech-to-speech / agent workflow
- Direction
- Higher is better
- Sample
- 246
- Source type
- community-reported
- Evaluator
- Cekura speech-to-speech agent benchmark
- Status
- active
- Retrieved at
- 2026-10-18
- Methodology
- Community-reported result. Cekura speech-to-speech whole-agent workflow comparison. Exact model ID gemini-3.8-live-extended-thinking (displayed as Gemini 3.8 Live Extended Thinking); thinking level high; voice Charon; service turn detection; audio 16 kHz. Both compared systems use the same open-source Pipecat pipeline, prompt, tools, and connection. 82 scenarios run three times each; the page reports 2,952 calls across the benchmark.
- Context
- This is an agent-workflow result, not a model-only score. The page gives 82 scenarios and three runs per model; no 95% CI or immutable serving revision is published. The evaluator's institutional independence and affiliations are not established by this page. Data accuracy is conditioned on calls that save data, with the subset denominator unknown. Results apply only to the exact agent configuration above. N=246 calls per model from 82 scenarios x 3 runs; call-based measure.
Cekura saved-data accuracy (Show measurement, test conditions, and source)
- Source value
- 84.4
- Score
- 84.4
- Metric
- calls where everything saved matched exactly
- Unit
- %
- Benchmark version
- Cekura speech-to-speech agent comparison, page last updated 2026-10-07
- Category
- speech-to-speech / agent workflow
- Direction
- Higher is better
- Source type
- community-reported
- Evaluator
- Cekura speech-to-speech agent benchmark
- Status
- active
- Retrieved at
- 2026-10-18
- Methodology
- Community-reported result. Cekura speech-to-speech whole-agent workflow comparison. Exact model ID gemini-3.8-live-extended-thinking (displayed as Gemini 3.8 Live Extended Thinking); thinking level high; voice Charon; service turn detection; audio 16 kHz. Both compared systems use the same open-source Pipecat pipeline, prompt, tools, and connection. 82 scenarios run three times each; the page reports 2,952 calls across the benchmark.
- Context
- This is an agent-workflow result, not a model-only score. The page gives 82 scenarios and three runs per model; no 95% CI or immutable serving revision is published. The evaluator's institutional independence and affiliations are not established by this page. Data accuracy is conditioned on calls that save data, with the subset denominator unknown. Results apply only to the exact agent configuration above. Unknown. The page conditions this measure on calls that save data but gives no denominator for that subset.
Page 1 of 4
Popularity compared
How often the model is searched for, used and mentioned, compared with every model of its category.
Popularity index
0 to 100, higher is more popular
20 of 83 models
pyannote Community-1, pyannote: 73
Nemotron-3-Diarization, NVIDIA: 57
Gemini 3.5 Transcribe, Google: 52
Confucius4-R2T2, NetEase Youdao: 52
Granite Speech 5.0 470M TurboCTC, IBM: 52
Kokoro-82M, hexgrad: 51
Nova-3, Deepgram: 51
Fish Audio S2 Pro, Fish Audio: 51
NVIDIA Nemotron 3.5 Transcribe, NVIDIA: 50
Granite Speech 5.0 470M TurboCTC NC, IBM: 50
Orukeet, Oruk: 49
Lyria 3.5, Google: 47
Gemini 3.5 Live Translate, Google: 46
Phonon-2, Fermion Research: 44
Aura-2, Deepgram: 44
Gemini 3.1 Flash TTS, Google: 44
MiniMax-Music3, MiniMax: 41
Audio8-ASR-Infinite, Edge0 AI: 41
GPT-Realtime 2.1, OpenAI: 41
Gemini 3.8 Live Extended Thinking, Google: 0
Gemini 3.8 Live Extended Thinking
As of November 2026 · Source: gradually.ai popularity index
Over time
Weekly values from October 18, 2026 to November 5, 2026
Gemini 3.8 Live Extended ThinkingAmazon Nova 2 SonicEleven Multilingual STS v2Eleven Music v2.5Eleven Text to Sound v2
All specifications
Architecture
Input and output
- Capabilities
- Audio-to-audio dialogue with interleaved reasoning, asynchronous function calling, and audio streaming
- Source: Capabilities
Watermarking and provenance
AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.
- AI watermarking
- Yes. Google documents SynthID for audio outputs of the listed Gemini Audio models.
- Source: AI watermarking
Licence and access
API
Further details
Prices
- Text input price
- $0.75 per 1M text tokensSource
- Audio input price
- $3 per 1M audio tokens ($0.005 per audio minute)Source
- Text output price
- $4.50 per 1M text tokensSource
- Audio output price
- $12 per 1M audio tokens ($0.018 per audio minute)Source
- Pricing basis
- Paid Standard tier, prices in USD. Per-minute rates are estimates based on token usage, not fixed per-minute charges.Source
Sources and data date
Show all 28 sources
Benchmark sources at a glance
Which independent evaluators tested this model. This is coverage, not a quality ranking.
No active result from the reviewed external evaluators can currently be attributed to this model. Other source evidence remains available below.
Without vendor claims and historical or disputed results; those stay labelled in the single values.