GPT-5.6 Sol vs Kimi K3 for property and lettings admin
GPT-5.6 Sol wins, 93–90
Same 12 property and lettings admin tasks, marked blind by three rival labs. GPT-5.6 Sol took 5 tasks, Kimi K3 took 2, 5 tied. Tested 18 Aug 2026.
Where they differed most
Messy notes from a portfolio handover, exactly as written: '14 Ashworth Rd, Leeds LS6 - gas cert done 3rd March 26, EPC band C, deposit £1150 with TDS, ref TDS-88421, EICR I think 2023 but no paperwork. Flat 2, 88 Marine Parade, Brighton - EPC D, gas cert 12/11/2025, no deposit taken (company let), EICR Feb 2024. 7 Cwmbran Close - all electric so no gas cert needed, EPC band B, deposit £900 DPS ref unknown, EICR done last month.' Reply with ONLY valid JSON — no prose before or after. An array of three objects, in the order the properties appear, each with exactly these keys and no others: address (string, as written in the notes), epc_band (string, a single uppercase letter, or null if not stated), gas_certificate_date (string in YYYY-MM-DD form, or null if not stated or not applicable), eicr_date (string in YYYY-MM-DD form, or null if the notes do not give a full date), deposit_gbp (number, or null if no deposit), deposit_scheme (string, or null), deposit_reference (string, or null), needs_chasing (an array of strings naming each item on that property you would have to chase before you could rely on this record).
[
{
"address": "14 Ashworth Rd, Leeds LS6",
"epc_band": "C",
"gas_certificate_date": "2026-03-03",
"eicr_date": null,
"deposit_gbp": 1150,
"deposit_scheme": "TDS",
"deposit_reference": "TDS-88421",
"needs_chasing": [
"EICR full date",
"EICR paperwork"
]
},
{
"address": "Flat 2, 88 Marine Parade, Brighton",
"epc_band": "D",
"gas_certificate_date": "2025-11-12",
"eicr_date": "2024-02-01",
"deposit_gbp": null,
"deposit_scheme": null,
"deposit_reference": null,
"n…[
{
"address": "14 Ashworth Rd, Leeds LS6",
"epc_band": "C",
"gas_certificate_date": "2026-03-03",
"eicr_date": null,
"deposit_gbp": 1150,
"deposit_scheme": "TDS",
"deposit_reference": "TDS-88421",
"needs_chasing": ["EICR date/report"]
},
{
"address": "Flat 2, 88 Marine Parade, Brighton",
"epc_band": "D",
"gas_certificate_date": "2025-11-12",
"eicr_date": null,
"deposit_gbp": null,
"deposit_scheme": null,
"deposit_reference": null,
"needs_chasing": ["EICR date/report"]
},…Task by task
| Task | GPT-5.6 Sol | Kimi K3 |
|---|---|---|
| Two routes, one worried tenant | 10 | 9 |
| Six weeks late and hoping | 9 | 9 |
| Write only the arrears email | 10 | 10 |
| The pre-tenancy checklist | 10 | 10 |
| Write only the renewal reply | 10 | 10 |
| Three repairs, one Tuesday | 10 | 8 |
| The nine-year-old carpet | 8 | 8 |
| The rent rise by text message | 10 | 8 |
| Every claim in the advert must be true | 8 | 9 |
| The compliance data, as JSON | 6 | 10 |
| What the agent must not promise | 10 | 8 |
| Refuse the lock change | 10 | 9 |
Full receipts: GPT-5.6 Sol, Kimi K3 · judges claude-sonnet-5, gemini-3.1-pro-preview, grok-4.5
Questions people ask
Which is better for property and lettings admin: GPT-5.6 Sol or Kimi K3?
GPT-5.6 Sol — it scored 93/100 against 90/100 on our 12-task property and lettings admin suite, winning 5 tasks to 2 with 5 tied. Every answer was marked blind by three judges from three rival AI labs.
How was this tested?
Both models answered the identical published property and lettings admin tasks. Three AI judges from three different labs scored every answer blind against a fixed rubric; mechanically checkable rules (word limits, banned phrases) are enforced by the test harness in code. The raw outputs and judge verdicts are downloadable.
More property and lettings admin head-to-heads: GPT-5.5 vs GPT-5.6 Sol · GPT-5.5 vs Kimi K3 · Claude Opus 4.8 vs GPT-5.6 Sol · Claude Fable 5 vs GPT-5.6 Sol · Claude Sonnet 5 vs GPT-5.6 Sol · GPT-5.6 Sol vs Grok 4.5
Full ranking: Best AI for property and lettings admin · model pages: GPT-5.6 Sol, Kimi K3