Claude Sonnet 5.5
| Maker | Anthropic |
|---|---|
| Family | Claude Sonnet |
| Released | 2026-09-28 |
| Context | 1M |
| Max output | 128K |
| Open weights | Closed |
|---|---|
| Modalities | text, image, pdf → text |
| Input / 1M | $2.00 |
| Output / 1M | $10.00 |
| Cached input / 1M | $0.10 |
| Review call | $0.0766 |
|---|---|
| Reasoning efforts | low, medium, high, xhigh, max |
| Hosts | Anthropic, OpenRouter |
| Model id | claude-sonnet-5-5 |
- Rank
- #1=
- of 17
- Bugs caught
- 77%
- 64%–89%
- False alarms
- 2.0
- per clean PR
- Cost / review
- $0.0571
- Cost / bug
- $0.083
- Time
- 11s
- per review
Claude Sonnet 5.5 is a fast first seat: a review came back in 11s at $0.0571, with 77% [64%–89%] recall. It ranks 1 of 17 and is measurably above 4 ranked models.
Against its pricier sibling the trade is clean. Claude Opus 5.5 cost 2.7× as much per review, took 33s, and measurably beats 9 ranked models, though the benchmark can't separate the two on recall. GPT-6.1 Sol lists at $2.00 / $10.00 per 1M tokens and Sonnet at $2.00 / $10.00, but a Sonnet review cost 1.4× Sol's. A Sol review took 28s, and again the recall gap isn't measurable.
The weak point is noise. Sonnet raised 2.0 [1.5–2.4] false alarms per clean PR. GPT-6 Luna and Gemini 3.1 Pro both averaged fewer, with intervals that don't overlap Sonnet's. If your team ignores noisy bots, pair Sonnet with a quieter seat or read reducing false positives.
Its misses include "Fix cache file length". The results post has the first look.
One note on scoring: an Anthropic judge is among the three, and its vote is dropped for this model.
Figures follow the latest benchmark run and price check. Text checked Nov 5, 2026.
| Host | Model id | Input / 1M | Output / 1M |
|---|---|---|---|
| Anthropic | claude-sonnet-5-5 | $2.00 | $10.00 |
Source: Anthropic pricing, Oct 16, 2026
By kind of bug
- Logic 14
- 64%
- Security 9
- 89%
- Concurrency 8
- 83%
- Error handling 5
- 80%
- API misuse 3
- 60%
- Resource handling 3
- 100%
Bug by bug
- Known bugs
- 42
- Caught every run
- 31
- Caught sometimes
- 2
- Missed
- 9
- Clean PRs
- 15
- False alarms on them
- 31
| Result | Pull request | Language | Kind |
|---|---|---|---|
| Caught sometimes4/5 | 4.x: Enhance Streamable/Streamer and DisposableContainer APIsReactiveX/RxJava | Java | API misuse |
| Caught sometimes2/3 | Refactor: MudDebouncedInput, ThrottleDispatcher and DebounceDispatcherMudBlazor/MudBlazor | C# | Concurrency |
| Missed | fix(query-core): fix hydration bugs for already resolved promisesTanStack/query | TypeScript | Logic |
| Missed | keep websocket compression state across interleaved control framesaio-libs/aiohttp | Python | Logic |
| Missed | Fix collection filtering API for IN/NOT IN comparisons that require type conversionsdoctrine/orm | PHP | Logic |
| Missed | fix(serialization): cap codec allocationsdotnet/orleans | C# | Logic |
| Missed | perf(gitdiff): optimize inline diff highlighting using cachego-gitea/gitea | Go | Error handling |
| Missed | refactor: Improve renaming/injection of resources in stripped APKsiBotPeaches/Apktool | Java | Security |
| Missed | Store cached child els in Attributes, not fieldjhy/jsoup | Java | Concurrency |
| Missed | KTOR-9171 Redesign ByteReadChannel.readUTF8Line APIktorio/ktor | Kotlin | Logic |
| Missed | Add the option to define custom cipher suites for backend serversTransporttraefik/traefik | Go | API misuse |
| Version | Released | Input / 1M | Output / 1M | Context |
|---|---|---|---|---|
| Claude Sonnet 5 | 2026-06-30 | $2.00 | $10.00 | 1M |
| Claude Sonnet 4.6 | 2026-02-17 | $3.00 | $15.00 | 1M |
| Model | Maker | Input / 1M | Output / 1M | Review call | Context | Bugs caught |
|---|---|---|---|---|---|---|
| GPT-6.1 Sol | OpenAI | $2.00 | $10.00 | $0.0766 | 1.1M | 89% |
| Gemini 4 Argon | $2.00 | $10.00 | $0.0766 | 1M | 82% | |
| Kimi K2.7 Code Highspeed | Moonshot | $1.90 | $8.00 | $0.0691 | 262.1K | – |
| Grok 4.7 | xAI | $2.00 | $6.00 | $0.0669 | 500K | – |
Updated Nov 5, 2026