Claude Opus 5.5 vs DeepSeek V4 Flash (Workers AI)
Reading a 26.2K-token pull request costs $0.1048 on Claude Opus 5.5, 9.1× the $0.0115 it costs on DeepSeek V4 Flash (Workers AI). Claude Opus 5.5 is a frontier model and DeepSeek V4 Flash (Workers AI) a economy one. In the benchmark Claude Opus 5.5 caught 41 percentage points more of the known defects than DeepSeek V4 Flash (Workers AI) (95% interval +29 pts to +53 pts, a − b).
Side by side
| Figure | Claude Opus 5.5 | DeepSeek V4 Flash (Workers AI) |
|---|---|---|
| Price tier | Frontier | Economy |
| Bugs caughtshare of known bugs | 90% (better) | 49% |
| False alarmsper clean PR | 1.7 | 1.3 (better) |
| Cost per reviewmeasured | $0.1525 | $0.0178 (better) |
| Cost per bug caught | $0.1921 | $0.0309 (better) |
| Exact linecatches on the expected lines | 89% (better) | 88% |
| Review timemean | 33s (better) | 137s |
| Typical review calllist price | $0.1532 | $0.0147 (better) |
| Input / 1M tokens | $4.00 | $0.44 (better) |
| Output / 1M tokens | $20.00 | $1.32 (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 Opus 5.5 | 90% | 100% | 50% | 78% | 100% | 100% | 100% | 67% | 100% | 100% | 89% | 100% | 100% | 100% |
| DeepSeek V4 Flash (Workers AI) | 49% | 46% | 25% | 55% | 46% | 22% | 56% | 61% | 78% | 61% | 44% | 67% | 50% | 0% |
Of the 42 known bugs both were scored on, both caught 29 and neither caught 2. Claude Opus 5.5 caught 10 bugs that DeepSeek V4 Flash (Workers AI) missed: 3 in TypeScript, 2 in C#, 1 in C, 1 in C++, 1 in JavaScript, 1 in Ruby and 1 in Rust. DeepSeek V4 Flash (Workers AI) caught one bug that Claude Opus 5.5 missed: 1 in Kotlin.
On pull requests with no known bug, Claude Opus 5.5 raised 1.7 false alarms per review and DeepSeek V4 Flash (Workers AI) 1.3. A measured review cost $0.1525 on Claude Opus 5.5, 8.6× the cost of $0.0178 on DeepSeek V4 Flash (Workers AI).
| Pull request | Language · kind | Claude Opus 5.5 | DeepSeek V4 Flash (Workers AI) |
|---|---|---|---|
| OrchardCMS/OrchardCoreSimplify using `StringValues` in Liquid templates | C# · Security | 1/1 | 0/3 |
| TanStack/queryfix(query-core): fix hydration bugs for already resolved promises | TypeScript · Logic | 1/1 | 0/4 |
| 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 | 1/1 | 0/3 |
| colinhacks/zodfeat(v4): parse input containing reference cycles | TypeScript · Resource handling | 1/1 | 0/4 |
| dotnet/orleansfix(serialization): cap codec allocations | C# · Logic | 1/1 | 0/3 |
| ggml-org/llama.cpprpc : cache and reuse compute graphs | C++ · Security | 1/1 | 0/3 |
| nushell/nushellreorganize nushell configuration | Rust · Error handling | 1/1 | 0/4 |
| sindresorhus/p-throttleAdd `weight` option | JavaScript · Logic | 1/1 | 0/3 |
| trpc/trpcfix(server): pass maxDurationMs timeout signal to subscription handlers | TypeScript · Concurrency | 1/1 | 0/1 |
| rails/railsReimplement Action Cable redis adapter with redis-client | Ruby · Logic | 2/3 | 0/3 |
| kotest/kotestFix DataClassEq comparator | Kotlin · Logic | 0/1 | 1/3 |
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: Claude Opus 5.5 and DeepSeek V4 Flash (Workers AI).