DeepSeek V4 Pro (Workers AI) vs GLM-5.3 Flash (Workers AI)
Reading a 26.2K-token pull request costs $0.0346 on DeepSeek V4 Pro (Workers AI), 8.8× the $0.0039 it costs on GLM-5.3 Flash (Workers AI). DeepSeek V4 Pro (Workers AI) is a mid-range model and GLM-5.3 Flash (Workers AI) a economy one. The benchmark cannot separate their recall: DeepSeek V4 Pro (Workers AI) minus GLM-5.3 Flash (Workers AI) is −7 pts (95% interval −20 pts to +6 pts).
Side by side
| Figure | DeepSeek V4 Pro (Workers AI) | GLM-5.3 Flash (Workers AI) |
|---|---|---|
| Price tier | Mid-range | Economy |
| Bugs caughtshare of known bugs | 54% | 61% (better) |
| False alarmsper clean PR | 2.9 | 1.8 (better) |
| Cost per reviewmeasured | $0.0643 | $0.0035 (better) |
| Cost per bug caught | $0.0872 | $0.0066 (better) |
| Exact linecatches on the expected lines | 89% | 89% |
| Review timemean | 390s | 254s (better) |
| Typical review calllist price | $0.0442 | $0.0051 (better) |
| Input / 1M tokens | $1.32 | $0.15 (better) |
| Output / 1M tokens | $3.96 | $0.50 (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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4 Pro (Workers AI) | 54% | 79% | 25% | 49% | 54% | 59% | 17% | 30% | 69% | 72% | 67% | 75% | 38% | 33% |
| GLM-5.3 Flash (Workers AI) | 61% | 57% | 31% | 45% | 88% | 30% | 67% | 89% | 33% | 78% | 67% | 67% | 100% | 67% |
Of the 42 known bugs both were scored on, both caught 31 and neither caught 4. DeepSeek V4 Pro (Workers AI) caught 3 bugs that GLM-5.3 Flash (Workers AI) missed: 2 in PHP and 1 in TypeScript, most of them logic bugs. GLM-5.3 Flash (Workers AI) caught 4 bugs that DeepSeek V4 Pro (Workers AI) missed: 2 in JavaScript, 1 in Go and 1 in Rust.
On pull requests with no known bug, DeepSeek V4 Pro (Workers AI) raised 2.9 false alarms per review and GLM-5.3 Flash (Workers AI) 1.8. A measured review cost $0.0643 on DeepSeek V4 Pro (Workers AI), 18× the cost of $0.0035 on GLM-5.3 Flash (Workers AI).
| Pull request | Language · kind | DeepSeek V4 Pro (Workers AI) | GLM-5.3 Flash (Workers AI) |
|---|---|---|---|
| TanStack/queryfix(query-core): fix hydration bugs for already resolved promises | TypeScript · Logic | 3/4 | 0/4 |
| doctrine/ormFix collection filtering API for IN/NOT IN comparisons that require type conversions | PHP · Logic | 3/4 | 0/3 |
| symfony/symfony[Security] Allow multiple OIDC discovery endpoints | PHP · Security | 1/3 | 0/3 |
| denoland/denoperf(libs/core): poll only ready handles and yield between I/O batches | Rust · Concurrency | 0/4 | 3/3 |
| golang/gocrypto/tls: add LocalCertificate to ConnectionState | Go · Logic | 0/4 | 3/4 |
| pinojs/thread-streamrefactor: replace setTimeout with Atomics.waitAsync | JavaScript · Concurrency | 0/4 | 1/3 |
| sindresorhus/p-throttleAdd `weight` option | JavaScript · Logic | 0/4 | 2/3 |
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 Pro (Workers AI) and GLM-5.3 Flash (Workers AI).