DeepSeek V4 Flash (Workers AI) vs Gemini 3.8 Flash
DeepSeek V4 Flash (Workers AI) reads a 26.2K-token pull request for $0.0115, 41% less than Gemini 3.8 Flash ($0.0197). DeepSeek V4 Flash (Workers AI) is a economy model and Gemini 3.8 Flash a mid-range one. The benchmark cannot separate their recall: DeepSeek V4 Flash (Workers AI) minus Gemini 3.8 Flash is −19 pts (95% interval −35 pts to −4 pts).
Side by side
| Figure | DeepSeek V4 Flash (Workers AI) | Gemini 3.8 Flash |
|---|---|---|
| Price tier | Economy | Mid-range |
| Bugs caughtshare of known bugs | 49% | 68% (better) |
| False alarmsper clean PR | 1.3 | 1.0 (better) |
| Cost per reviewmeasured | $0.0178 (better) | $0.0862 |
| Cost per bug caught | $0.0309 (better) | $0.1166 |
| Exact linecatches on the expected lines | 88% | 94% (better) |
| Review timemean | 137s (better) | 427s |
| Typical review calllist price | $0.0147 (better) | $0.0287 |
| Input / 1M tokens | $0.44 (better) | $0.75 |
| Output / 1M tokens | $1.32 (better) | $3.75 |
| Context window | 1M | 1M |
Head to head
The known bugs both were scored on, by who caught each in at least one run.
By language
| Model | All | TypeScript6 bugs | Go4 bugs | Java4 bugs | Rust4 bugs | C#3 bugs | JavaScript3 bugs | Kotlin3 bugs | PHP3 bugs | Python3 bugs | Ruby3 bugs | Swift3 bugs | C++2 bugs | C1 bug |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4 Flash (Workers AI) | 49% | 46% | 25% | 55% | 46% | 22% | 56% | 61% | 78% | 61% | 44% | 67% | 50% | 0% |
| Gemini 3.8 Flash | 68% | 83% | 75% | 43% | 75% | 78% | 67% | 0% | 50% | 83% | 100% | 100% | 50% | 50% |
Of the 42 known bugs both were scored on, both caught 23 and neither caught 4. DeepSeek V4 Flash (Workers AI) caught 7 bugs that Gemini 3.8 Flash missed: 3 in Kotlin, 2 in Java, 1 in JavaScript and 1 in PHP. Gemini 3.8 Flash caught 8 bugs that DeepSeek V4 Flash (Workers AI) missed: 2 in C#, 2 in TypeScript, 1 in C, 1 in Go, 1 in JavaScript and 1 in Ruby.
On pull requests with no known bug, DeepSeek V4 Flash (Workers AI) raised 1.3 false alarms per review and Gemini 3.8 Flash 1. A measured review cost $0.0178 on DeepSeek V4 Flash (Workers AI), 79% less than $0.0862 on Gemini 3.8 Flash.
| Pull request | Language · kind | DeepSeek V4 Flash (Workers AI) | Gemini 3.8 Flash |
|---|---|---|---|
| kotest/kotestRefactor eventually to ensure wall-clock time usage | Kotlin · Error handling | 3/3 | 0/1 |
| nodejs/undicifeat: add configurable maxPayloadSize for WebSocket | JavaScript · Security | 2/3 | 0/1 |
| ktorio/ktorKTOR-9171 Redesign ByteReadChannel.readUTF8Line API | Kotlin · Logic | 3/6 | 0/1 |
| jhy/jsoupStore cached child els in Attributes, not field | Java · Concurrency | 2/6 | 0/4 |
| kotest/kotestFix DataClassEq comparator | Kotlin · Logic | 1/3 | 0/1 |
| symfony/symfony[Security] Allow multiple OIDC discovery endpoints | PHP · Security | 1/3 | 0/1 |
| mockito/mockitoAdd core API to enable Kotlin singleton mocking | Java · API misuse | 1/4 | 0/1 |
| OrchardCMS/OrchardCoreSimplify using `StringValues` in Liquid templates | C# · Security | 0/3 | 1/1 |
| TanStack/queryfix(query-core): fix hydration bugs for already resolved promises | TypeScript · Logic | 0/4 | 1/1 |
| c-ares/c-aresEvent Thread could stall when not notified of new queries on existing connections that are in a bad state | C · Resource handling | 0/3 | 1/2 |
| dotnet/orleansfix(serialization): cap codec allocations | C# · Logic | 0/3 | 1/1 |
| go-chi/chifeat: middleware.ClientIP, a replacement for middleware.RealIP | Go · Security | 0/3 | 1/1 |
| rails/railsReimplement Action Cable redis adapter with redis-client | Ruby · Logic | 0/3 | 1/1 |
| sindresorhus/p-throttleAdd `weight` option | JavaScript · Logic | 0/3 | 1/1 |
| trpc/trpcfix(server): pass maxDurationMs timeout signal to subscription handlers | TypeScript · Concurrency | 0/1 | 1/1 |
Defects caught over all runs: a PR reviewed three times with one known bug shows 3/3 when it was caught every time.
Every bug each was scored on: DeepSeek V4 Flash (Workers AI) and Gemini 3.8 Flash.