Claude Sonnet 5.5 vs DeepSeek V4 Pro (Workers AI)
DeepSeek V4 Pro (Workers AI) reads a 26.2K-token pull request for $0.0346, 34% less than Claude Sonnet 5.5 ($0.0524). Both sit in the mid-range price tier, so the choice between them is about review quality more than budget. The benchmark cannot separate their recall: Claude Sonnet 5.5 minus DeepSeek V4 Pro (Workers AI) is +23 pts (95% interval +7 pts to +39 pts).
Side by side
| Figure | Claude Sonnet 5.5 | DeepSeek V4 Pro (Workers AI) |
|---|---|---|
| Price tier | Mid-range | Mid-range |
| Bugs caughtshare of known bugs | 77% (better) | 54% |
| False alarmsper clean PR | 2.0 (better) | 2.9 |
| Cost per reviewmeasured | $0.0571 (better) | $0.0643 |
| Cost per bug caught | $0.083 (better) | $0.0872 |
| Exact linecatches on the expected lines | 82% | 89% (better) |
| Review timemean | 11s (better) | 390s |
| Typical review calllist price | $0.0766 | $0.0442 (better) |
| Input / 1M tokens | $2.00 | $1.32 (better) |
| Output / 1M tokens | $10.00 | $3.96 (better) |
| 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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Claude Sonnet 5.5 | 77% | 83% | 50% | 45% | 100% | 56% | 100% | 67% | 67% | 67% | 100% | 100% | 100% | 100% |
| DeepSeek V4 Pro (Workers AI) | 54% | 79% | 25% | 49% | 54% | 59% | 17% | 30% | 69% | 72% | 67% | 75% | 38% | 33% |
Of the 42 known bugs both were scored on, both caught 27 and neither caught 2. Claude Sonnet 5.5 caught 6 bugs that DeepSeek V4 Pro (Workers AI) missed: 2 in Go, 2 in JavaScript, 1 in Rust and 1 in TypeScript. DeepSeek V4 Pro (Workers AI) caught 7 bugs that Claude Sonnet 5.5 missed: 2 in Java, 1 in Go, 1 in Kotlin, 1 in PHP, 1 in Python and 1 in TypeScript, most of them logic bugs.
On pull requests with no known bug, Claude Sonnet 5.5 raised 2 false alarms per review and DeepSeek V4 Pro (Workers AI) 2.9. A measured review cost $0.0571 on Claude Sonnet 5.5, 11% less than $0.0643 on DeepSeek V4 Pro (Workers AI).
| Pull request | Language · kind | Claude Sonnet 5.5 | DeepSeek V4 Pro (Workers AI) |
|---|---|---|---|
| colinhacks/zodfeat(v4): parse input containing reference cycles | TypeScript · Resource handling | 1/1 | 0/4 |
| denoland/denoperf(libs/core): poll only ready handles and yield between I/O batches | Rust · Concurrency | 1/1 | 0/4 |
| go-chi/chifeat: middleware.ClientIP, a replacement for middleware.RealIP | Go · Security | 1/1 | 0/4 |
| golang/gocrypto/tls: add LocalCertificate to ConnectionState | Go · Logic | 1/1 | 0/4 |
| pinojs/thread-streamrefactor: replace setTimeout with Atomics.waitAsync | JavaScript · Concurrency | 1/1 | 0/4 |
| sindresorhus/p-throttleAdd `weight` option | JavaScript · Logic | 1/1 | 0/4 |
| TanStack/queryfix(query-core): fix hydration bugs for already resolved promises | TypeScript · Logic | 0/1 | 3/4 |
| aio-libs/aiohttpkeep websocket compression state across interleaved control frames | Python · Logic | 0/1 | 4/4 |
| doctrine/ormFix collection filtering API for IN/NOT IN comparisons that require type conversions | PHP · Logic | 0/1 | 3/4 |
| iBotPeaches/Apktoolrefactor: Improve renaming/injection of resources in stripped APKs | Java · Security | 0/1 | 2/3 |
| jhy/jsoupStore cached child els in Attributes, not field | Java · Concurrency | 0/2 | 3/6 |
| ktorio/ktorKTOR-9171 Redesign ByteReadChannel.readUTF8Line API | Kotlin · Logic | 0/1 | 1/7 |
| traefik/traefikAdd the option to define custom cipher suites for backend serversTransport | Go · API misuse | 0/1 | 4/4 |
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: Claude Sonnet 5.5 and DeepSeek V4 Pro (Workers AI).