Gemini 3.1 Pro vs Gemini 4 Argon
Gemini 3.1 Pro and Gemini 4 Argon charge practically the same to read a 26.2K-token pull request ($0.0524). Above 200K tokens long-context rates apply, and Gemini 4 Argon becomes the cheaper one. 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: Gemini 3.1 Pro minus Gemini 4 Argon is −15 pts (95% interval −28 pts to −3 pts).
Side by side
| Figure | Gemini 3.1 Pro | Gemini 4 Argon |
|---|---|---|
| Price tier | Mid-range | Mid-range |
| Bugs caughtshare of known bugs | 67% | 82% (better) |
| False alarmsper clean PR | 0.8 (better) | 1.1 |
| Cost per reviewmeasured | $0.2499 | $0.0527 (better) |
| Cost per bug caught | $0.4134 | $0.0706 (better) |
| Exact linecatches on the expected lines | 89% (better) | 78% |
| Review timemean | 349s | 99s (better) |
| Typical review calllist price | $0.0814 | $0.0766 (better) |
| Input / 1M tokens | $2.00 | $2.00 |
| Output / 1M tokens | $12.00 | $10.00 (better) |
| Context window | 1M | 1M |
- Above 200K tokens, long-context rates mean Gemini 4 Argon is the cheaper of the two.
- From 2026-12-01, after a scheduled reprice, Gemini 3.1 Pro is the cheaper of the two.
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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Gemini 3.1 Pro | 67% | 67% | 38% | 78% | 88% | 56% | 33% | 67% | 67% | 67% | 100% | 100% | 50% | 50% |
| Gemini 4 Argon | 82% | 83% | 50% | 83% | 75% | 89% | 67% | 100% | 100% | 83% | 100% | 100% | 100% | 0% |
Of the 42 known bugs both were scored on, both caught 29 and neither caught 3. Gemini 3.1 Pro caught 3 bugs that Gemini 4 Argon missed: 1 in C, 1 in Go and 1 in Rust. Gemini 4 Argon caught 7 bugs that Gemini 3.1 Pro missed: 1 in C++, 1 in C#, 1 in Go, 1 in JavaScript, 1 in Kotlin, 1 in PHP and 1 in TypeScript, most of them logic bugs.
On pull requests with no known bug, Gemini 3.1 Pro raised 0.8 false alarms per review and Gemini 4 Argon 1.1. A measured review cost $0.2499 on Gemini 3.1 Pro, 4.7× the cost of $0.0527 on Gemini 4 Argon.
| Pull request | Language · kind | Gemini 3.1 Pro | Gemini 4 Argon |
|---|---|---|---|
| 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 | 1/2 | 0/1 |
| golang/gocrypto/tls: add LocalCertificate to ConnectionState | Go · Logic | 1/2 | 0/1 |
| nushell/nushellreorganize nushell configuration | Rust · Error handling | 1/2 | 0/1 |
| TanStack/queryfix(query-core): fix hydration bugs for already resolved promises | TypeScript · Logic | 0/1 | 1/1 |
| dotnet/orleansfix(serialization): cap codec allocations | C# · Logic | 0/1 | 1/1 |
| ggml-org/llama.cpprpc : cache and reuse compute graphs | C++ · Security | 0/1 | 1/1 |
| go-chi/chifeat: middleware.ClientIP, a replacement for middleware.RealIP | Go · Security | 0/1 | 1/1 |
| kotest/kotestFix DataClassEq comparator | Kotlin · Logic | 0/1 | 1/1 |
| sindresorhus/p-throttleAdd `weight` option | JavaScript · Logic | 0/1 | 1/1 |
| symfony/symfony[Security] Allow multiple OIDC discovery endpoints | PHP · Security | 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: Gemini 3.1 Pro and Gemini 4 Argon.