Gemini 3.8 Flash
Low-cost Google model near the top for writing and maths; rankings are preliminary and API prices double in January.
Should you use it?
Pick it if
People who write a lot or work with maths, video or audio input and want a low per-token price.
Best for: Writing, maths and video or audio input at a low token price (Chat models Top 20)
Skip it if
You mainly build software; GPT-6.1 Sol and Claude Sonnet 5.5 rank far higher on WebDev.
You give up: Weak at web app coding (rank 31); API prices double on 1 Jan 2027
The short version
- What this is
- Google's main fast Gemini model, released 2 Sep 2026, in the Gemini app, Search AI Mode, AI Studio and the Gemini API.
- Why it matters
- It ranks near the top for writing and maths at a low per-token price, but its rankings are still preliminary.
- Seventh on arena.ai Creative Writing and fourth on Math, both preliminary s1 s3.
- API $0.75 / $3.75 per 1M tokens until the end of 2026, then double s9.
- In the Gemini app from Google AI Pro at $19.99/month s8.
Good at
Our take
On the public boards, Gemini 3.8 Flash is better at writing and maths than its price suggests: seventh on arena.ai Creative Writing and fourth on Math on 11 Oct 2026 s1 s3. Both rows are marked preliminary, so both could move as votes build.
Two cautions on value. It spends a lot of tokens: on the Artificial Analysis index it costs $1.24 a task, more than GPT-6.1 Sol at max effort s5. And Google's promotional API price ends on 31 Dec 2026, when input and output prices double s9. In the Gemini app it needs Google AI Pro at $19.99/month; the Free plan lists an older Flash-Lite model s8.
Scores by job
A score from 0 to 100 per job, against everything available today. Overall 77 across 4 jobs, the average of the job scores.
New to this? What the scores and labels mean
- 0 to 100
- Our score for one job, like writing or coding, compared with everything else out today. 100 is the best available today. For AI models we build it from public data, such as Arena's blind votes and Epoch AI's independent tests; apps and tools are scored on our own checklist. The same tool can score differently for different jobs, and where two likely ranges overlap, we call it a tie.
- Our verdict
- Use it: a top choice for most people. Worth trying: good, and worth a go. Depends: good for some people or one kind of job. Wait: too early to tell, or not out for everyone yet. Skip: something else does the job better.
- First look
- Just released. We say what it is and whether to bother, and give it a score once public results exist.
- No score yet
- None of the public sources we use covers it yet, or the only evidence is from a source whose numbers we can't use, so we describe it in words instead.
- Best buy
- Nothing cheaper scores as well for that job.
- Best free
- The best score for that job that costs nothing, through a free plan or by running on your own computer.
- Pay as you go
- No monthly plan: you pay for each use, usually through a developer account.
- Checked
- The day we last confirmed the score and the price against the sources. Prices change often, so check the date.
- AI model
- The engine underneath, like GPT or Claude. You use it through an app or, if you code, directly.
- Open source
- Free to download and run on your own computer, so nothing leaves your machine.
#3 of the 12 we have scored for this job.
From blind votes (Arena) · Arena data, CC BY 4.0
Early, few votesData read 20 Oct
#7 of the 15 we have scored for this job.
From blind votes (Arena) and independent tests (Epoch AI) · Arena data, CC BY 4.0, Epoch AI data, CC BY 4.0
Early, few votesData read 20 Oct
The top score of the 2 we have scored for this job.
From blind votes (Arena) and independent tests (Epoch AI) · Arena data, CC BY 4.0, Epoch AI data, CC BY 4.0
Early, few votesData read 20 Oct
#20 of the 21 we have scored for this job.
From blind votes (Arena) · Arena data, CC BY 4.0
Early, few votesData read 20 Oct
Alternatives
Better or cheaper for writing, from our scores, then the ones we'd also look at. The same kind of product comes first.
- Claude Opus 5.5BetterText model100$20 a monthScores 100 in writing, against 91 here; $20 a month. Side by side
- Gemini 4 ArgonCheaperText model99$4.99 a monthScores 99 in writing for $4.99 a month, against $19.99 a month here. Side by side
- Claude Sonnet 5.5FreeText model85FreeScores 85 in writing, with a free plan. Pick it when coding matters as much as writing. Side by side
- GPT-6.1 SolText model83$20 a monthMuch stronger at coding and agents, and cheaper per index task, though weaker at creative writing. Side by side
- Muse Spark 1.3Text model83$5 a monthSimilar price per token, ranks higher on WebDev and Agent. Side by side
Value for money
Where it sits in writing: higher is a better score, further left is cheaper. Green dots are the Best buys.
Evidence
Everything behind the call and the scores, folded so the page stays short. Spotted a wrong price, score or link? Report a mistake.
How we scored this
The writing, research, data and coding scores are our own, built from public data: blind votes on Arena and, where they cover the job, Epoch AI's independent tests. How we combine them is our method and we don't publish it.
Public leaderboardIndependent test
Reviews are drafted with help from AI models, including Anthropic's Claude. Model scores come only from outside leaderboards and independent evaluations, by a written rule.
No one paid for this review or gave us free access, and we earn nothing if you sign up. How we score
Sources 12
- s1arena.ai Text: Creative Writing leaderboard
- s2arena.ai Text: Instruction Following leaderboard
- s3arena.ai Text: Math leaderboard
- s4arena.ai Text: Expert leaderboard
- s5Artificial Analysis LLM leaderboard
- s6arena.ai Code Arena: WebDev leaderboard
- s7arena.ai Text: Coding leaderboard
- s8Google AI plans (Gemini subscriptions)
- s9Gemini API pricing
- s10arena.ai Agent leaderboard
- s11Gemini Apps Privacy Hub
- s12Google DeepMind: Gemini 3.8 Flash model card
Something here out of date or wrong? Report a mistake and we will check it against the source.
History 2 changes
Every change to a rating, the price or our call, with the reason. Scores are against what exists on the day, so they drift down as better things ship. Older entries come from before we moved to scores.
- New: Writing ratedNew: Data ratedNew: Research ratedNew: Coding ratedReviewed: writing, data, research, coding.
- Audit: Agent cost per task corrected from $0.59 to $0.61 (the main table's median for Gemini 3.8 Flash (High)); Text Coding second-source note corrected to the distinct-model rank. Ratings unchanged.
Use it well 4 guides on this site
Pages on this site that help with the jobs above.