Google: Gemini 3.8 Flash
Gemini 3.8 Flash is a reasoning model from Google released on Sep 2, 2026. It reads text, images, files, audio and video and writes text, with a 1.05M token context window and up to 66K tokens per reply.
Show moreShow less
It costs $0.75 per million input tokens and $3.75 per million output tokens. Cached input bills at $0.075, 90% less than fresh input.
It supports tool calling, structured JSON output and prompt caching.
24 providers serve it, including Vercel AI Gateway, Vertex and 302.AI.
On the Artificial Analysis Intelligence Index it scores 40.9, which places it at number 14 among the 43 models I track with independent scores.
- Modalities
- In / out price
- $0.75 / $3.75 per 1M
- Context
- 1.05M
- Max output
- 66K
- Released
- Sep 2, 2026
- Knowledge cutoff
- n/a
- Tool calling
- Structured output
- Reasoning
- Prompt caching
- Open weights
- Free to use
Gemini 3.8 Flash pricing
USD, as listed on Oct 15, 2026. Most bills are input heavy, so the cached input rate matters as much as the headline price.
| Input | $0.75 / 1M tokens | |
|---|---|---|
| Output | $3.75 / 1M tokens | |
| Cached input (read) | $0.075 / 1M tokens | 90% off input |
| Audio input | $0.75 / 1M tokens |
Gemini 3.8 Flash providers
24 providers serve it, including Vercel AI Gateway, Vertex and 302.AI. Each host sets its own price and limits, so the cheapest one is not always the one with the longest context.
| Provider | Input /M | Output /M | Cache read /M | Context | Max output |
|---|---|---|---|---|---|
| Google First party | $0.75 | $3.75 | $0.075 | 1.05M | 66K |
| Vercel AI Gateway | $0.75 | $3.75 | $0.075 | 1M | 66K |
| Vertex | $0.75 | $3.75 | $0.075 | 1.05M | 66K |
| 302.AI | $0.75 | $3.75 | – | 1.05M | 66K |
| AIHubMix | $0.75 | $3.75 | $0.075 | 1.05M | 66K |
| CrossModel | $0.75 | $3.75 | $0.075 | 1.05M | 66K |
| DevPass (LLM Gateway) | $0.75 | $3.75 | $0.075 | 1.05M | 1.05M |
| Kilo Gateway | $0.75 | $3.75 | $0.075 | 1.05M | 66K |
| LLM Gateway | $0.75 | $3.75 | $0.075 | 1.05M | 66K |
| Merge Gateway | $0.75 | $3.75 | $0.075 | 1.05M | 66K |
| NanoGPT | $0.75 | $3.75 | $0.075 | 1.05M | 66K |
| NEAR AI Cloud | $0.75 | $3.75 | $0.075 | 1.05M | 66K |
Showing 12 of 24 providers: the lab first, then the major clouds, then the cheapest resellers.
What Gemini 3.8 Flash costs per month
Pick a workload close to yours or type your own volumes.
Independent benchmarks
Measured by third parties, not reported by Google. Each links to its source and shows the date I checked it.
A composite of 10 independent evaluations (agents 30%, coding 20%, scientific reasoning 20%, general 30%), run by Artificial Analysis on each model, version 4.3.2.
Artificial Analysis, checked Oct 10, 2026
Human preference: people compare two anonymous answers and pick the better one. Ratings from the Text Arena overall board, style control on.
Arena (formerly LMArena), checked Oct 10, 2026
Agentic coding in a real terminal: can the model, driving an agent harness, finish multi step software and ops tasks. Resolution rate.
Terminal-Bench, checked Oct 10, 2026
Abstract reasoning on visual puzzles the model has not seen, semi private set. Hard to memorise, so it tracks reasoning more than knowledge.
ARC Prize, checked Oct 10, 2026
Expert level questions across many fields, no tools allowed. The Diamond set is 1,000 questions. Few models are on it yet.
Scale Labs, checked Oct 10, 2026
Median output tokens per second measured by Artificial Analysis on the provider API, for the same model setting as its index score.
Artificial Analysis, checked Oct 10, 2026
Benchmarks Google reports (13)
- deepswe-v1.1 (DeepSWE v1.1, long-horizon software engineering)73.7%
- gdpval-aa-v2 (GDPVal-AA v2, knowledge work)1545
- vals-finance-agent-v2 (Vals Finance Agent v2, financial analyst tasks)61.4%
- harvey-legal-agent (Harvey's Legal Agent Benchmark, complex legal workflows, all pass rate)10%
- terminal-bench-2.1 (Terminal-bench 2.1, agentic terminal coding (no harness stated on the 3.8 Flash card))89.4%
- terminal-bench-4.0 (Terminal-bench 4.0, general agent capabilities)19.1%
- gdp-pdf (GDP.PDF, expert PDF document comprehension, all pass rate)35%
- charxiv-reasoning (CharXiv Reasoning, no tools)86.2%
- lvbench (LVBench, long video understanding, agentic)87.8%
- hle-verified (HLE-Verified, multidisciplinary expert reasoning (tool setting not stated))54.9%
- osworld-2.0 (OSWorld-2.0, agentic computer use, partial score, batch tool enabled)59%
- biomysterybench (BioMysteryBench, human solvable)88.8%
- labbench2 (LABBench2, biology real-world research tasks)86.2%
Labs pick their own settings and harnesses, so compare these only within one lab.
Supported request parameters
What you can set when you call Gemini 3.8 Flash through an OpenAI compatible API.
- Tools
tools - Structured outputs
structured_outputs - Reasoning
reasoning - Temperature
temperature
Alternatives to Gemini 3.8 Flash
Models that take the same inputs and call tools, closest in price, cheaper ones first.
- 1.05M context
Reads text, images, files, audio and video. Tool calling, structured output, reasoning. Served by 25 providers.
by googleJul 9, 2026$0.75/M input tokens$3.75/M output tokens25 providers
- 1.05M context
Reads text, images, files, audio and video. Tool calling, structured output, reasoning. Served by 27 providers.
by googleDec 17, 2025$0.50/M input tokens$3/M output tokens27 providers
- 1.05M context
Reads text, images, files, audio and video. Tool calling, structured output, reasoning. Served by 24 providers.
by googleAug 13, 2026$0.75/M input tokens$3.75/M output tokens24 providers
- 1.05M context
Reads text, images, files, audio and video. Tool calling, structured output, reasoning. Served by 6 providers.
by googleAug 13, 2026$0.75/M input tokens$3.75/M output tokens6 providers
- Index 22.2
Reads text, images, files, audio and video. Tool calling, structured output, reasoning. Served by 24 providers.
by googleJul 9, 20261.05M context$0.30/M input tokens$2.50/M output tokens24 providers
- 1.05M context
Reads text, images, files, audio and video. Tool calling, structured output, reasoning. Served by 32 providers.
by googleMar 20, 2025$0.30/M input tokens$2.50/M output tokens32 providers
More from Google
Nano Banana 2.1
Image66K contextGenerates images from text, images and files. Served by 4 providers.
by googleOct 6, 2026$1.50/M input tokens$7.50/M output tokens4 providers
- 1.05M context
Reads text, images, files, audio and video. Tool calling, structured output, reasoning. Served by 13 providers.
by googleSep 30, 2026$2/M input tokens$10/M output tokens13 providers
Turns text into spoken audio.
by googleSep 23, 2026$0.50/M input tokens$6/M output tokens
Gemini 3.8 Flash TTS
SpeechTurns text into spoken audio.
by googleSep 23, 2026$0.50/M input tokens$9/M output tokens
- 262K context
Reads text. Reasoning.
by googleSep 19, 2026$0.05/M input tokens$0.15/M output tokens
- 262K context
Reads text and images. Tool calling, reasoning.
by googleSep 19, 2026$0.1056/M input tokens$0.3344/M output tokens
Gemini 3.8 Flash questions
How much does the Gemini 3.8 Flash API cost?
$0.75 per million input tokens and $3.75 per million output tokens, with cached input at $0.075. The estimator on this page turns that into a monthly bill for your own workload.
What is the context window of Gemini 3.8 Flash?
1,048,576 tokens of input, and it can write up to 65,536 tokens in one reply. As a rough guide, a million tokens is around 750,000 English words.
Which providers serve Gemini 3.8 Flash?
24 providers serve it, including Vercel AI Gateway, Vertex and 302.AI. Prices differ by host, so the Providers table on this page lists each one with its input and output price.
Does Gemini 3.8 Flash support tool calling?
Yes. It accepts tool definitions and can return JSON that matches a schema, which is what an agent needs to call your APIs.
Is Gemini 3.8 Flash open source?
No. Google serves it through an API only, so you cannot download the weights or host it yourself.
When was Gemini 3.8 Flash released?
Sep 2, 2026.
How good is Gemini 3.8 Flash compared with other models?
It scores 40.9 on the Artificial Analysis Intelligence Index and 1497 on the Arena text leaderboard. Benchmarks are a shortlist, not a verdict: test the top two or three candidates on your own tasks before you commit.
Catalog data from models.dev (MIT licensed), refreshed daily, last on Oct 15, 2026. Independent scores from my hand checked rankings, sources linked above. The summary is written from these facts.
Building on Gemini 3.8 Flash?
I test your real tasks on it and the two closest alternatives, then ship the agent on whichever wins on cost and accuracy.
Book Meeting