Skip to content
Tier A — Frontier
Runs in:USMade in:United States
Google Gemini

Gemini 3.8 Flash

Tier A — Frontier · 1.048576M tokens

Tokonomix Editorial Team·Reviewed by Mes Kalkan··
Section 01

Speed analysis

Latency measured across all benchmark runs. P50 (median) and P95 (95th percentile) give a realistic picture of response speed under normal and peak load.

P50 latency (median)P95 latency107 runs
10643338561378871016110-2611-22ms
Section 02

Quality scores

How this model compares to the rest of the field on each prompt category, from a pairwise fit over the same prompts. The raw judge score sits underneath each number.

29%
Coding
judge mean 47
10%
Creative
judge mean 45
46%
Factual
judge mean 63
25%
Reasoning
judge mean 45

Win rate per category: how often this model beats a field-average model on a prompt from that category. 50% is average, not a failing grade. It is not a percentage of correct answers.

Section 03

Pricing

What you pay per million tokens when you use this model on Tokonomix, plus an estimate for a typical conversation.

💰
API rates — Gemini 3.8 Flash
$1.13 per 1M input tokens
$5.61 per 1M output tokens
≈ $0.0018 per typical conversation (800 tokens)
Input vs output price (per 1M tokens)
per 1M input tokens$1.13
per 1M output tokens$5.61
Section 04

Tokens per second

Throughput in tokens per second, derived from measured P50 latency. Higher is better; fluctuations track provider-side load.

Throughput (tokens / s)101 / avg 119
18658

Estimated from P50 latency × 200 output tokens — the absolute number depends on this assumption; the trend is what matters.

Section 05

Capabilities

toolssource: seed-newest-generation-2026-09-30visionjson modepdf inputreasoningaudio inputjson schemaparallel toolsprompt cachingoutputTokenLimit: 65536max output tokens: 65535
Section 06

Availability

Availability

How often this model answers when we call it — measured across real API requests and live tests over the last 30 days. This is separate from quality: these numbers only tell you whether the model responds, not how good the answer is.

Last 7 days

100.0%

n=6

Last 30 days

100.0%

n=183

Median response time

5,436ms

n=183

Based on 557 measurements over the last 30 days.

Technical details

Only live API calls and live-test requests count — internal probes and benchmark runs are excluded.

Calls with a custom API key (BYOK) are excluded: those failures are key-specific, not a sign of model downtime.

Failed calls are NOT included in quality scores — quality is measured on successful responses only. Availability and quality are independent signals.

Median response time (p50) across successful calls with a recorded duration. Outliers (very slow or very fast calls) pull the median less than the average.

Total calls (30d)

183

OK responses (30d)

183

Total calls (7d)

6

OK responses (7d)

6

Section 07

Tokonomix benchmark verdicts

⚖️
Endorsed by 1 judge
Independent LLM judges evaluated this model on our weekly intelligence tests
claude-sonnet-4-551/100 · 25 runs
10 correct0 partial15 wrong40% accuracy
● 2026-11-22

Gemini 3.8 Flash debuts with multimodal tools and expanded input support

Gemini 3.8 Flash enters benchmarking as a new model with comprehensive multimodal capabilities. The model supports vision, audio input, and PDF processing alongside standard text capabilities. It includes both single and parallel tool calling with JSON mode and schema support, plus reasoning features and prompt caching for optimization. Without previous performance data, this represents Google's latest flash-tier offering with a full feature set designed for multimodal applications. The capabilities package positions it as a versatile model for developers needing document understanding, image analysis, and audio processing in a single API. Tool support with parallel execution enables complex agentic workflows. Prompt caching availability suggests optimization for applications with repeated context. The model appears aimed at production use cases requiring multiple input modalities and structured outputs. Users evaluating this model should test it against their specific multimodal workloads, particularly scenarios combining vision, audio, and document processing with tool use. Performance characteristics, latency profiles, and output quality will need real-world validation as benchmark data accumulates over time.

Quality

—

Latency p50

—

Test runs

0

✓ Full multimodal input support added✓ Parallel tools and JSON schema✓ Prompt caching now available✓ PDF and audio processing enabled
Last automated test
Nov 22, 2026 · 14:02 UTC · Speed benchmark
P50 latency
1978 ms
P95 latency
3506 ms
Errors
0 / 6 runs
Last reviewed by Tokonomix Team·November 22, 2026