Gemini 3.8 Flash vs DeepSeek V4.1 Flash (1017)
Gemini 3.8 Flash and DeepSeek V4.1 Flash (1017) are within three points on the Models at Work leaderboard, which we treat as a tie (65 vs 62 out of 100). On independent tests alone, Gemini 3.8 Flash is ahead (63 vs 43). DeepSeek V4.1 Flash (1017) is cheaper: 1,000 support replies cost about $1.38 at list price, against $3.75 on Gemini 3.8 Flash, which is 2.7 times as much. On the Artificial Analysis Intelligence Index, one of the independent tests, DeepSeek V4.1 Flash (1017) leads 33 to 31.
| Gemini 3.8 Flash · Google | DeepSeek V4.1 Flash (1017) · DeepSeek | |
|---|---|---|
| Leaderboard rank | #12 | #17 |
| Score (0–100) | 65 | 62 |
| Kind of work | High-volume work | High-volume work |
| Price per 1M tokens | $0.75 to read · $3.75 to write | $0.30 to read · $1.20 to write |
| Reads at once | 1M tokens (about 750,000 words) | 1M tokens (about 750,000 words) |
| Where you can get it | Gemini Enterprise Agent Platform (Google Cloud), Gemini app (Google AI Pro and Ultra) | DeepSeek API, Gemini Enterprise Agent Platform (Google Cloud), Microsoft Foundry |
| Can a company host it itself | No | Yes (open weights) |
| Artificial Analysis Intelligence Index | 31 | 33 |
| Arena Text score | 1497 (rank 8) | Not measured |
| Scale SWE-Bench Pro V2 (mini-swe-agent) | 58.80% | Not measured |
| Humanity's Last Exam, Diamond (Scale) | 34.30 | 20.50 |
| Humanity's Sixth Sense (Scale) | 41.60 | 30.10 |
| Best for | First-pass support replies, Summarising documents, images and recordings at volume, Routine multi-step tasks on a budget | Bulk drafting and rewriting, First-pass replies at volume, Trials where data location is not a constraint |
Cost of 1,000 jobs
| 1,000 of these | Gemini 3.8 Flash | DeepSeek V4.1 Flash (1017) |
|---|---|---|
| Support repliesthe customer message, the account notes and two help articles read; a 300-word reply written | $3.75 | $1.38 |
| Documents processeda two-page invoice or form read; the fields pulled out | $2.63 | $0.96 |
| Calls summariseda 40-minute call transcript read; a summary and next steps written | $9.75 | $3.72 |
| Contracts revieweda 40-page contract read against a checklist; a memo of issues written | $28 | $11 |
| Drafts writtena brief and one example read; a 600-word email or proposal section written | $4.13 | $1.41 |
Worked out from each vendor’s list price and a fixed size for each job. A product built on a model charges its own price on top. See what every model costs for the same jobs.
Choose Gemini 3.8 Flash for
- First-pass support replies
- Summarising documents, images and recordings at volume
- Routine multi-step tasks on a budget
Fast and well liked in blind votes for the money, at under half the list price of Gemini 4 Argon, but users report more made-up answers than rivals and Google doubles the price on 1 January 2027.
Caveats: $0.75 in / $3.75 out per million tokens (the unit vendors bill by) is introductory; Google's pricing page says $1.50 / $7.50 from 1 January 2027. Users report more invented facts and more dropped details between messages than with rival models: have someone check its output. Third Flash release in six weeks: expect Google to retire versions quickly and plan for re-testing. In the Gemini app it is for Google AI Pro and Ultra subscribers, per Google's launch post. Google Workspace is not listed: no Google page we read names the model and a Workspace plan.
Choose DeepSeek V4.1 Flash (1017) for
- Bulk drafting and rewriting
- First-pass replies at volume
- Trials where data location is not a constraint
About a seventh of Kimi K3.1 Pro's list price for 33 on Artificial Analysis's index, and now sold by Google's and Microsoft's clouds as well as DeepSeek itself.
Caveats: DeepSeek's own service is in China and charges more at peak hours: $0.15 to $0.30 in and $0.60 to $1.20 out per million tokens. Read its data terms, or buy through a US cloud. Artificial Analysis rebuilt its index this month and the score fell from 45 to 33; on Scale's two hard exams it sits well below the top models. The weights are free to use commercially (MIT licence), per DeepSeek. DeepSeek's own test figures are not counted.
What people running them say
Gemini 3.8 Flash: A very large Hacker News (developer forum) thread: 1,160 upvotes, 669 comments. simonw had it build an interactive chart as a web page in '13 seconds' for '1.8 cents'; colechristensen calls it 'competitive with opus/fable and also FAST'; several use it for the bulk of the work while a stronger model writes the plan. The pushback is about reliability: one user says it 'gives me the most hallucinations' (made-up facts) of the major models, others complain it ignores what was said one message earlier, and in some subjects its knowledge stops at early 2025, which hurts research tasks.
DeepSeek V4.1 Flash (1017): A 1,067-upvote thread on Hacker News, a developer forum, about the earlier Flash release said the low price is real and so is the gap to Opus at writing software; nothing newer was read this run, so this is unscored.
Questions people ask
Which is better, Gemini 3.8 Flash or DeepSeek V4.1 Flash (1017)?
Gemini 3.8 Flash and DeepSeek V4.1 Flash (1017) are within three points on the Models at Work leaderboard, which we treat as a tie (65 vs 62 out of 100). On independent tests alone, Gemini 3.8 Flash is ahead (63 vs 43). DeepSeek V4.1 Flash (1017) is cheaper: 1,000 support replies cost about $1.38 at list price, against $3.75 on Gemini 3.8 Flash, which is 2.7 times as much. On the Artificial Analysis Intelligence Index, one of the independent tests, DeepSeek V4.1 Flash (1017) leads 33 to 31.
When should I choose Gemini 3.8 Flash over DeepSeek V4.1 Flash (1017)?
Choose Gemini 3.8 Flash for first-pass support replies, summarising documents, images and recordings at volume, routine multi-step tasks on a budget. Fast and well liked in blind votes for the money, at under half the list price of Gemini 4 Argon, but users report more made-up answers than rivals and Google doubles the price on 1 January 2027.
When should I choose DeepSeek V4.1 Flash (1017) over Gemini 3.8 Flash?
Choose DeepSeek V4.1 Flash (1017) for bulk drafting and rewriting, first-pass replies at volume, trials where data location is not a constraint. About a seventh of Kimi K3.1 Pro's list price for 33 on Artificial Analysis's index, and now sold by Google's and Microsoft's clouds as well as DeepSeek itself.
Which is cheaper, Gemini 3.8 Flash or DeepSeek V4.1 Flash (1017)?
Gemini 3.8 Flash costs $0.75 per million tokens it reads (input) and $3.75 per million tokens it writes (output) on Google's list price (Gemini API Pricing, read Nov 12, 2026). A token is about three-quarters of a word. DeepSeek V4.1 Flash (1017) costs $0.30 per million tokens it reads (input) and $1.20 per million tokens it writes (output) on DeepSeek's list price (DeepSeek API pricing, read Nov 12, 2026). A token is about three-quarters of a word.
What do 1,000 jobs cost on Gemini 3.8 Flash and DeepSeek V4.1 Flash (1017)?
At list price: 1,000 support replies cost $3.75 on Gemini 3.8 Flash and $1.38 on DeepSeek V4.1 Flash (1017); 1,000 documents processed cost $2.63 on Gemini 3.8 Flash and $0.96 on DeepSeek V4.1 Flash (1017); 1,000 calls summarised cost $9.75 on Gemini 3.8 Flash and $3.72 on DeepSeek V4.1 Flash (1017); 1,000 contracts reviewed cost $28 on Gemini 3.8 Flash and $11 on DeepSeek V4.1 Flash (1017); 1,000 drafts written cost $4.13 on Gemini 3.8 Flash and $1.41 on DeepSeek V4.1 Flash (1017). A product built on either model charges its own price on top.
Scores come from the Models at Work leaderboard: Score is 0 to 100 and computed, not typed. 40% is independent testing (Artificial Analysis, Arena, Scale, Epoch, Vals and others; each test is scaled against the best score on this board, and tests of real work, such as finished freelance projects or professional tasks, count double the exams). 30% is price: the vendor's list price for a typical mix of reading and writing, on a fixed scale from $0.05 to $60 per million tokens, so a model that costs a fifth as much gains real ground. 15% is what people report after running it, from Signals and linked threads. 15% is fit: how much text it can take in at once and where a business can get it (the products and clouds that carry it), with a small credit for open weights. A missing component drops out and the rest are reweighted. A model with no independent test yet is provisional and ranks below every measured one. Within three points is a tie. Vendor-published figures are labelled and never counted. Prices are vendor list prices where published; a typical mix assumes three parts reading to one part writing.