Ratings / Anthropic
Claude Haiku 5.5
Anthropic · claude-haiku-5-5 · effort medium (default)
Haiku 5.5withAPIRanked #4 overall
46 runs · ±3.0 · Solid evidenceWhat you can hand it, and what you’ll still have to catch
Dropped 8 Oct 2026, against Claude Haiku 4.5 and whatever leads today.
Nearly Sonnet 5.5's judgement. Watch it with numbers & calculations.
Haiku 5.5 has a Combined score of 84. That's just three points behind Sonnet 5.5 and four behind GPT-6.1 Sol, at a twentieth of Sonnet 5.5's list price. The LLM judge found 29 of its 44 graded written outputs usable with at most a quick edit, and none made a critical mistake. It's strongest when the job is a judgement call, like why retention is flat, whether to launch, which OKRs to keep. It slips when it has to do the maths itself or write a full spec. Both of its PRDs needed rework.
What you can hand it
- Retention diagnoses. Both scored 96. It found the real cause in the data and gave a plan you could act on with light polish.
- OKRs. It cut eleven key results down to a set a team could own, and both outputs were ready to send after a quick edit.
- Launch calls. It made a clear call region by region against the agreed criteria, with the conditions that would change it.
- Challenging an idea. Its review of the AI SDR plan got a perfect score. It went straight for the assumption the plan rests on and kept to the word limit.
- Performance calibration. Its pre-read on fourteen ratings from three managers also got a perfect score, with specific changes and the evidence behind each one.
- 1000x an idea. On both cases it found what the data was pointing at instead of just scaling the idea up.
What you’ll still have to catch
- Numbers it computes. It said total MRR was $11,670 when it's $11,820, put a week-one lift at 4.0% when it's 5.6%, and gave a 66-day sales cycle where the data says 62. Re-check any figure it calculated rather than copied.
- Facts that aren't in the brief. It gave a rival 2,100 partner firms, assumed a "participant session log" the platform doesn't have, and stated "search traffic compounds" as fact.
- PRDs. Neither was usable with a quick edit. Its tests had no pass mark or measurement window, and it never said who the feature isn't for.
- Growth loop maths. It named the loops but didn't work out what each one yields once you account for retention.
- Discovery that doesn't commit. On the screen-time interviews it listed possibilities without ranking them, and it said "every participant" and "several parents" where it should have counted.
- Best setupHaiku 5.5 · API84.1#4 overall
- Against the leader−5.9GPT-6 Astra · ChatGPT
Task by task
Its Combined score on each task type, across 46 outputs.
| Model | Overall | Product Craft | Growth | Metrics & Experimentation | Strategy | Leadership | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Discovery | Research guide | PRD | Roadmap | Prototype | Launch call | Activation | Growth loop | GTM plan | Retention | Test design | Test readout | North Star | OKRs | Strategy | Challenge | 1000x | Rivals | PMF call | Pricing | Exec update | Hiring | Reviews | ||
| GPT-6 AstraOpenAI | 90#1 | 96Best | 81 | 84Best | 97 | 70 | 89 | 91Best | 84 | 100Best | 98Best | 79 | 98Best | 91 | 99Best | 87Best | 88 | 92 | 93 | 96Best | 86 | 94Best | 86 | 91Best |
| Haiku 5.5Anthropic | 84#4 | 77 | 77 | 67 | 86 | 92 | 92 | 82 | 70 | 82 | 96 | 89 | 78 | 86 | 96 | 80 | 92 | 93Best | 80 | 81 | 76 | 89 | 79 | 91 |
Combined scoreEach cell is the model’s best setup on that task type.
Every tested harness
Same model, different setup. The harness often matters more than the launch post says.
| Rank | Harness | Combined score | Decision model | LLM judge | Evidence |
|---|---|---|---|---|---|
| 4 | Haiku 5.5withAPIOpenRouter 2026-09 | 85.3 | 83.9 |
Published runs
46 outputs, each graded blind on its task’s current checklist
| Task type · task | Harness | Combined score |
|---|---|---|
| 1000x an ideaBudgets for two | API | |
| 1000x an ideaSharing a shopping list, 1000x | API | |
| Activation & onboarding reviewThe checklist isn't the moment | API | |
| Activation & onboarding reviewAnalytics tool losing users at setup | API | |
| Analyse experiment resultsUp 6%, and significant | API | |
| Analyse experiment resultsThe underpowered onboarding test | API | |
| Build a roadmapA year of spend management, with a hard deadline | API | |
| Build a roadmapFourteen requests from one customer | API | |
| Call product-market fit38%, and the retention that isn't | API | |
| Call product-market fitDo we have fit, or a spike? | API | |
| Challenge an ideaFive billion dollars, on paper | API | |
| Challenge an ideaAn AI SDR for small agencies | API | |
| Competitor analysisFree forever, funded | API | |
| Competitor analysisThe competitor isn't who you think | API | |
| Customer research call guideToo expensive, by default | API | |
| Customer research call guideWhy are warehouses leaving? Two execs, two theories | API | |
| Develop product strategySupply or demand for a stalled marketplace | API | |
| Develop product strategyEnterprise, says the CEO | API | |
| Diagnose a retention problemThe collapse that wasn't | API | |
| Diagnose a retention problemFlat, while sign-ups grow | API | |
| Draft OKRsA great quarter, on paper | API | |
| Draft OKRsEleven key results and a bonus | API | |
| Experiment specificationShow the fee sooner | API | |
| Experiment specificationShowing the delivery fee up front | API | |
| Extract discovery insightsHidden validation case | API | |
| Extract discovery insightsThe loudest ticket | API | |
| Find the growth loopThe badge on every form | API | |
| Find the growth loopGrowing on the surface, decaying underneath | API | |
| Hire a PMWhat our interviews predict | API | |
| Hire a PMA loop for the first growth PM | API | |
| Make the launch callGo/no-go for AI substitutions, from the rollout data | API | |
| Make the launch callOffline, on older phones | API | |
| One-shot prototypeExpense receipt capture | API | |
| One-shot prototypeClinic appointment rebooking | API | |
| Pick a North Star metricRevenue, actives, or something else? | API | |
| Pick a North Star metricGrowing, on paper | API | |
| Price and package a productFifteen dollars for everyone | API | |
| Price and package a productUnlimited pages, falling margins | API | |
| Write a GTM planEnterprise, without killing the engine | API | |
| Write a GTM planPayroll, for the people who fear it | API | |
| Write a performance reviewShipped a lot, moved nothing | API | |
| Write a performance reviewFourteen ratings, three managers | API | |
| Write a PRDMeeting summaries with action items | API | |
| Write a PRDDisputes that aren't disputes | API | |
| Write a stakeholder updateOn track, mostly | API | |
| Write a stakeholder updateA quarterly business review deck, in brand | API |