Gemini 3.8 Flash
Google · released Sep 2, 2026
- Type
- Closed
- Input
- $0.75
- Output
- $3.75
- Cached
- None held
List price · per 1M tokens · Google AI at 1M context · machine-readable source ↗
Our take
Written Sep 29, 2026Gemini 3.8 Flash is a hosted-only model with no download and no licence terms, so using it means choosing a host. It leads the LiveBench field on instruction following and places near the top of the Arena boards for maths and creative writing, but sits near the bottom on data analysis and below the middle on code.
Pick it for constrained rewriting and rule-following work, where it leads the LiveBench field, or for maths and creative writing where human preference is the measure. It also takes text, images, files, audio and video in one request, so a recording or a document does not have to be described first. Skip it if the job is data analysis or code generation, or if you need to run the model yourself.
The case for it
- 81.41% on LiveBench Instruction Following, 1st of 58 on that board as of 25 Jun 2026 — the board measures constrained-rewriting tasks, so it is evidence for rule-following rather than open-ended quality.
- 6th of 168 on Arena Creative Writing via High as of 25 Sep 2026, a human pairwise vote that records which answer people preferred rather than whether it was correct.
- Text, image, file, audio and video input in one request, so a recording or a document does not have to be transcribed or described first.
- The request capacity leaves room for a long report or a stack of documents beside the question, though reliable recall across all of it is unverified in our data.
The case against it
- 54.01% on LiveBench Data Analysis, 57th of 58 on that board as of 25 Jun 2026 — the board measures table and event-ordering tasks, so it is a poor fit for structured-data work.
- 72.49% on LiveBench Coding, 46th of 58 on that board as of 25 Jun 2026, and 24th of 95 on Arena Code (WebDev) via High as of 25 Sep 2026.
- We list no download for it, so using it means choosing a host, and no licence is supplied, so nothing here states what you may do with it.
How good is it?
A general-purpose assistant for everyday questions, drafting and coding, with agent work as a further strength.
- getting answers to everyday questionsArena Text (overall) · 8th of 168
- drafts, rewrites and editingArena Creative Writing · 6th of 168
- writing and completing codeArena Coding · 12th of 168
- multi-step work it carries out for youArena Agent · 12th of 55
EverydayGeneral questions and everyday reasoning
Arena Text (overall)8th of 168 · 1492
CodingWriting and fixing code on its own
Arena Coding12th of 168 · 1532
AgenticPlanning, calling tools, staying on task
Arena Agent12th of 55 · 0.041
WritingDrafting and rewriting prose
Arena Creative Writing6th of 168 · 1482
Placings on Arena's agent boards, from live sessions people ran themselves. A model can lead on one of these and sit mid-field on the others.
Boards this model appears on that none of the ratings above are built on.
Every published score for this model20 scoresEvery figure we hold, from 20 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Where to rent it
Prices checked 3 hours ago — each listing carries its own date.
Google AI, direct
The lab is the cheapest at this context. The strip above and this offer are the same one, compared at 1M of context. 2 cheaper rows below are outside that comparison: a non-standard pricing tier.
- per 1M tokens
- $0.75 in / $3.75 out
- Context served
- 1M
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| Google AI Studioflex tierThrough OpenRouter | $0.38 / $1.88checked 3 hours ago | 1M66K max reply | 144 tok/s | No | Yes55 days | Unknown |
| Google Vertex AIflex tierglobalThrough OpenRouter | $0.38 / $1.88checked 3 hours ago | 1M66K max reply | 25 tok/s | No | No | Confirmed |
| Google Vertex AIglobalThrough OpenRouter | $0.75 / $3.75checked 3 hours ago | 1M66K max reply | 64 tok/s | No | No | Confirmed |
| OpenRouterOpenRouter's own listing | $0.75 / $3.75checked 3 hours ago | 1M | not measured | Unknown | Unknown | Unknown |
| Google AI StudioThrough OpenRouter | $0.75 / $3.75checked 3 hours ago | 1M66K max reply | 129 tok/s | No | Yes55 days | Unknown |
| Google AIDirect | $0.75 / $3.75checked 3 hours ago | 1M66K max reply | not measured | Unknown | Unknown | Unknown |
| Google Vertex AIpriority tierglobalThrough OpenRouter | $1.35 / $6.75checked 3 hours ago | 1M66K max reply | 101 tok/s | No | No | Confirmed |
| Google AI Studiopriority tierThrough OpenRouter | $1.35 / $6.75checked 3 hours ago | 1M66K max reply | 152 tok/s | No | Yes55 days | Unknown |
Google AI Studio
Google Vertex AI
OpenRouterAcross the 8 listings we hold: 6 say they do not train on prompts, 0 say they do and 2 do not say. 3 appear in the zero-retention registry we check; the rest are unknown to us.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| Google AI StudioflexThrough OpenRouter | ✓ | ✓ | ✓ |
| Google Vertex AIflex · globalThrough OpenRouter | ✓ | ✓ | ✓ |
| Google Vertex AIglobalThrough OpenRouter | ✓ | ✓ | ✓ |
| OpenRouterOpenRouter's own listing | ✓ | ✓ | ✓ |
| Google AI StudioThrough OpenRouter | ✓ | ✓ | ✓ |
| Google AIDirect | |||
| Google Vertex AIpriority · globalThrough OpenRouter | ✓ | ✓ | ✓ |
| Google AI StudiopriorityThrough OpenRouter | ✓ | ✓ | ✓ |
Tool calling: 7 of 8 listings say yes, 1 publishes no parameter list. JSON output: 7 of 8 listings say yes, 1 publishes no parameter list. Strict schema: 7 of 8 listings say yes, 1 publishes no parameter list.
Models people weigh against Gemini 3.8 Flash
When we formed this view
Recent changes
Each date is the day we first saw the change, or the day the maker announced it.
What we do not know about this model yet
- 1 of 8 listings publishes no parameter list, so what its API accepts is unknown to us.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 2 of 8 listings do not say whether they train on prompts.
- We hold no batch or off-peak rate for any of its listings.
Licence and identifiers
What the licence allowsWe hold no licence record for this model. Inside are the identifiers you need to pull it — its Hugging Face repo where we have one, our slug and a machine-readable card.
Licence
We hold no licence record for this model, and no record of published weights either — so we can neither summarise its terms nor point you at the weights.
Identifiers
- Takes in, gives back
- Text, images, audio, video and documents in, text out
- Catalogue slug
- google-gemini-3-8-flash