GLM-5.3 (Workers AI) vs GPT-6 Luna
Reading a 26.2K-token pull request costs $0.0367 on GLM-5.3 (Workers AI), 14× the $0.0026 it costs on GPT-6 Luna. GLM-5.3 (Workers AI) is a mid-range model and GPT-6 Luna a economy one. The benchmark cannot separate their recall: GLM-5.3 (Workers AI) minus GPT-6 Luna is −6 pts (95% interval −21 pts to +9 pts).
Side by side
| Figure | GLM-5.3 (Workers AI) | GPT-6 Luna |
|---|---|---|
| Price tier | Mid-range | Economy |
| Bugs caughtshare of known bugs | 61% | 67% (better) |
| False alarmsper clean PR | 1.9 | 0.7 (better) |
| Cost per reviewmeasured | $0.0311 | $0.0028 (better) |
| Cost per bug caught | $0.0559 | $0.0046 (better) |
| Exact linecatches on the expected lines | 85% | 97% (better) |
| Review timemean | 65s | 29s (better) |
| Typical review calllist price | $0.0473 | $0.0038 (better) |
| Input / 1M tokens | $1.40 | $0.10 (better) |
| Output / 1M tokens | $4.40 | $0.50 (better) |
| 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-6 Luna | 67% | 83% | 25% | 65% | 75% | 56% | 67% | 67% | 67% | 100% | 67% | 100% | 50% | 0% |
Of the 42 known bugs both were scored on, both caught 27 and neither caught 4. GLM-5.3 (Workers AI) caught 9 bugs that GPT-6 Luna missed: 2 in Go, 1 in C, 1 in C++, 1 in Java, 1 in Kotlin, 1 in Ruby, 1 in Rust and 1 in TypeScript. GPT-6 Luna caught 2 bugs that GLM-5.3 (Workers AI) missed: 1 in Go and 1 in Ruby.
On pull requests with no known bug, GLM-5.3 (Workers AI) raised 1.9 false alarms per review and GPT-6 Luna 0.7. A measured review cost $0.0311 on GLM-5.3 (Workers AI), 11× the cost of $0.0028 on GPT-6 Luna.
| Pull request | Language · kind | GLM-5.3 (Workers AI) | GPT-6 Luna |
|---|---|---|---|
| ggml-org/llama.cpprpc : cache and reuse compute graphs | C++ · Security | 3/3 | 0/1 |
| socketry/asyncBarrier wait in completion order + `Queue#close`. | Ruby · Concurrency | 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 |
| traefik/traefikAdd the option to define custom cipher suites for backend serversTransport | Go · API misuse | 2/4 | 0/1 |
| jhy/jsoupStore cached child els in Attributes, not field | Java · Concurrency | 2/6 | 0/2 |
| golang/gocrypto/tls: add LocalCertificate to ConnectionState | Go · Logic | 1/4 | 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 |
| rails/railsReimplement Action Cable redis adapter with redis-client | Ruby · Logic | 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-6 Luna.