GLM-5.3 (Workers AI) vs GPT-5.6 Sol
Reading a 26.2K-token pull request costs $0.1048 on GPT-5.6 Sol, 2.9× the $0.0367 it costs on GLM-5.3 (Workers AI). GLM-5.3 (Workers AI) is a mid-range model and GPT-5.6 Sol a frontier one. The benchmark cannot separate their recall: GLM-5.3 (Workers AI) minus GPT-5.6 Sol is −12 pts (95% interval −26 pts to +4 pts).
Side by side
| Figure | GLM-5.3 (Workers AI) | GPT-5.6 Sol |
|---|---|---|
| Price tier | Mid-range | Frontier |
| Bugs caughtshare of known bugs | 61% | 72% (better) |
| False alarmsper clean PR | 1.9 | 1.2 (better) |
| Cost per reviewmeasured | $0.0311 (better) | $0.1127 |
| Cost per bug caught | $0.0559 (better) | $0.1732 |
| Exact linecatches on the expected lines | 85% | 92% (better) |
| Review timemean | 65s | 50s (better) |
| Typical review calllist price | $0.0473 (better) | $0.1532 |
| Input / 1M tokens | $1.40 (better) | $4.00 |
| Output / 1M tokens | $4.40 (better) | $20.00 |
| Context window | 1M | 1.1M |
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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GLM-5.3 (Workers AI) | 61% | 78% | 19% | 46% | 73% | 48% | 33% | 72% | 44% | 58% | 67% | 100% | 100% | 67% |
| GPT-5.6 Sol | 72% | 67% | 50% | 69% | 75% | 56% | 100% | 67% | 100% | 100% | 67% | 100% | 50% | 0% |
Of the 42 known bugs both were scored on, both caught 29 and neither caught 3. GLM-5.3 (Workers AI) caught 7 bugs that GPT-5.6 Sol missed: 2 in TypeScript, 1 in C, 1 in C++, 1 in Go, 1 in Kotlin and 1 in Rust. GPT-5.6 Sol caught 3 bugs that GLM-5.3 (Workers AI) missed: 1 in Go, 1 in JavaScript and 1 in PHP, most of them security bugs.
On pull requests with no known bug, GLM-5.3 (Workers AI) raised 1.9 false alarms per review and GPT-5.6 Sol 1.2. A measured review cost $0.0311 on GLM-5.3 (Workers AI), 72% less than $0.1127 on GPT-5.6 Sol.
| Pull request | Language · kind | GLM-5.3 (Workers AI) | GPT-5.6 Sol |
|---|---|---|---|
| ggml-org/llama.cpprpc : cache and reuse compute graphs | C++ · Security | 3/3 | 0/1 |
| withastro/astroPort Go's csrf protection | TypeScript · Logic | 4/4 | 0/1 |
| 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 | 2/3 | 0/1 |
| kotest/kotestFix DataClassEq comparator | Kotlin · Logic | 2/3 | 0/1 |
| trpc/trpcfix(server): pass maxDurationMs timeout signal to subscription handlers | TypeScript · Concurrency | 2/3 | 0/1 |
| traefik/traefikAdd the option to define custom cipher suites for backend serversTransport | Go · API misuse | 2/4 | 0/1 |
| nushell/nushellreorganize nushell configuration | Rust · Error handling | 1/4 | 0/1 |
| go-chi/chifeat: middleware.ClientIP, a replacement for middleware.RealIP | Go · Security | 0/3 | 1/1 |
| sindresorhus/p-throttleAdd `weight` option | JavaScript · Logic | 0/3 | 1/1 |
| symfony/symfony[Security] Allow multiple OIDC discovery endpoints | PHP · Security | 0/3 | 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: GLM-5.3 (Workers AI) and GPT-5.6 Sol.