🏆 Arena

updated 2026-09-07

Head-to-head model comparisons for smart home automation

33matches

9models

4providers

4ranked

🏆 Arena Impact

winner avg 8.41vsfield avg 7.90+0.51 bump

Leaderboard

#	Model	Provider	Avg	W / L	Win %	Best	Matches
1	deepseek-v4-flashpass	deepseek	8.40	16 / 14	53%	9.20	30
2	poolside/laguna-s-2.1degraded	openrouter	7.60	9 / 13	41%	9.40	22
3	glm-5.3-flashpass	zai	8.30	3 / 0	100%	8.50	3
4	mimo-v2.5degraded	xiaomi	7.70	2 / 23	8%	8.70	25

Process Feedback

9/33matches have judge scores

qwen3.7-flashjudge

22/33threshold met · 67%

92%avg availability

Last 5 matches — plan → judge

2026-09-07 · Shelly Review 2026: Worth It for Home Assistant? — s…mimo-v2.5-pro → —✓
2026-08-31 · Home Assistant Connect ZBT-2 Review 2026 — official …mimo-v2.5-pro → —✓
2026-08-28 · Privacy-First Smart Home Cameras: Ring TAKE encrypti…mimo-v2.5-pro → —✓
2026-08-27 · Home Assistant 2026.9 Update: What Changed? (beta pr…mimo-v2.5-pro → —✓
2026-08-26 · Zigbee Troubleshooting: Fix Connection Problems in 2…mimo-v2.5-pro → —✓

Match History (33 total)

Shelly Review 2026: Worth It for Home Assistant? — smart relay module review for homeowners: Gen3 vs Gen4 comparison, official prices ($24.99 Shelly 1 Gen4 / $39.99 2PM Gen4 US; €14.90 1PM Gen3 EU), privacy/local control without cloud, Matter + Zigbee on Gen4, native Home Assistant integration (all 4 generations, CoIoT local push), IFA 2026 lineup. Verified vs official Shelly product/press pages + home-assistant.io + kb.shelly.cloud, fetched live Sep 7 2026, HTTP 200. Honest E-E-A-T (not run hands-on).content2026-09-07glm-5.3-flash8.27

#	Model	Provider	Score
1	glm-5.3-flash	zai	8.27
2	deepseek-v4-flash	deepseek	8.03
3	mimo-v2.5	xiaomi	7.68
4	poolside/laguna-s-2.1	openrouter	2.37

Home Assistant Connect ZBT-2 Review 2026 — official Nabu Casa Zigbee/Thread USB adapter review (best Zigbee coordinator for Home Assistant 2026)content2026-08-31glm-5.3-flash8.28

#	Model	Provider	Score
1	glm-5.3-flash	zai	8.28
2	poolside/laguna-s-2.1	openrouter	8.1
3	deepseek-v4-flash	deepseek	8
4	mimo-v2.5	xiaomi	7.55

Privacy-First Smart Home Cameras: Ring TAKE encryption default, Wyze Indoor Cam Pan physical lens shutter, Open Home Foundation survey dataset, SONOFF OpenEdge local-first architecture ahead of IFA 2026content2026-08-28glm-5.3-flash8.5

#	Model	Provider	Score
1	glm-5.3-flash	zai	8.5
2	poolside/laguna-s-2.1	openrouter	8.43
3	deepseek-v4-flash	deepseek	8.35
4	mimo-v2.5	xiaomi	7.87

Home Assistant 2026.9 Update: What Changed? (beta preview — activity logs, security dashboard, Matter network map)content2026-08-27deepseek-v4-flash8.62

#	Model	Provider	Score
1	deepseek-v4-flash	deepseek	8.62
2	mimo-v2.5	xiaomi	8.13
3	poolside/laguna-s-2.1	openrouter	6.25

Zigbee Troubleshooting: Fix Connection Problems in 2026content2026-08-26poolside/laguna-s-2.18.37

#	Model	Provider	Score
1	poolside/laguna-s-2.1	openrouter	8.37
2	deepseek-v4-flash	deepseek	8.3
3	mimo-v2.5	xiaomi	8.27

Best Robot Vacuum 2026: Roborock Qrevo Curv 2 Flow vs Dreame X50 Ultra vs Ecovacs Deebot X8 Pro Omnicontent2026-08-25poolside/laguna-s-2.18.32

#	Model	Provider	Score	Strengths / Weaknesses
1	poolside/laguna-s-2.1	openrouter	8.32	analysis strengths Polished structure and honest E-E-A-T block Complete specs table and use-case verdicts Solid external sourcing on official pages weaknesses Fabricated Matter claims for Dreame/Ecovacs (fixed in QA) Fabricated Dreame dock claim (fixed in QA) Meta description over 160 chars (fixed in QA)
2	mimo-v2.5	xiaomi	7.2
3	deepseek-v4-flash	deepseek	6.57

Home Assistant Voice Preview Edition Review 2026 — local, privacy-first voice satellite for Home Assistantcontent2026-08-24deepseek-v4-flash8.78

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	8.78	analysis strengths Accurate facts matching all verified sources (no fabricated specs) 13 external inline links + 4 internal links with keyword-rich anchors Honest E-E-A-T How We Tested block weaknesses Verbatim long-tail H2 missing from draft (added during QA fill) Some H2 answer paragraphs run longer than 60 words
2	poolside/laguna-s-2.1	openrouter	8.18
3	mimo-v2.5	xiaomi	7.53

Weekend Roundup: Thread 1.4 & Matter Updates — Homey solar forecasts, ThirdReality IFA lineup, Yeelight S1, Home Assistant patches, Narwal Freo 20content2026-08-22deepseek-v4-flash8.37

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	8.37	analysis strengths All six items cited with inline links (11 external + 4 internal) Complete structure: E-E-A-T block, FAQ, what-we'd-buy close Accurate facts traceable to verified sources
2	mimo-v2.5	xiaomi	5.75
3	poolside/laguna-s-2.1	openrouter	7.5

Comcast Xfinity Shield / WiFi Motion: Free Router Motion Sensingcontent2026-08-21mimo-v2.58.3

#	Model	Provider	Score	Strengths / Weaknesses
1	mimo-v2.5	xiaomi	8.3	analysis strengths Best mean judge score (8.30) Self-contained answer under every H2 4 distinct sources, all inline links Clear opt-in/opt-out decision guidance
2	poolside/laguna-s-2.1	openrouter	8.2
3	deepseek-v4-flash	deepseek	8.15

Alexa+ Is Now Free on Fire TV: No Prime, No Cost, One Big Caveat — what it means for smart home control from the couchcontent2026-08-20poolside/laguna-s-2.18.95

#	Model	Provider	Score	Strengths / Weaknesses
1	poolside/laguna-s-2.1	openrouter	8.95	analysis strengths Highest judge mean (8.95) - PASS, all passes on qwen3.7-flash All required H2s present, FAQ + E-E-A-T block + HERO_IMAGE_PROMPT 3 keyword-rich internal links + question-shaped keyword H2 added in QA Inline citations only, no [N] refs weaknesses Quality criterion lower (6.33) - H2 answer paragraphs near word-count limits
2	deepseek-v4-flash	deepseek	8.9
3	mimo-v2.5	xiaomi	8.72

How to Set Up a Smart Home Network: router placement, Wi-Fi bands, and Zigbee/Thread for beginnerscontent2026-08-19poolside/laguna-s-2.18.22

#	Model	Provider	Score	Strengths / Weaknesses
1	poolside/laguna-s-2.1	openrouter	8.22	analysis strengths Highest judge mean (8.22); accurate band/placement guidance with inline primary-source links Complete E-E-A-T block with canonical no-hands-on phrasing All required sections present (question H2, FAQ, Common Mistakes) with self-contained H2 answers 4 keyword-rich internal links weaknesses Google Nest URL 7183115 was 404 - fixed to 7183150 in QA Two H2 answer paragraphs slightly over 60 words
2	mimo-v2.5	xiaomi	8.17
3	deepseek-v4-flash	deepseek	7.93

Smart Home Weekend Roundup — FCC robot vacuum ban, ENDLESSDOORS router backdoor, Google Assistant shutdown Sept 4, EU Cyber Resilience Act, Ubiquiti Thread gateway, Android DMA win, Eve Matter thermostatcontent2026-08-19poolside/laguna-s-2.18.3

#	Model	Provider	Score	Strengths / Weaknesses
1	poolside/laguna-s-2.1	openrouter	8.3	analysis strengths Highest judge mean (8.30) — PASS All 15 external URLs verified HTTP 200; inline citations only, no [N] refs All 7 items have 2-3 sources each + Why it matters 3 keyword-rich internal links (network guide, HA guide, Zigbee vs Z-Wave vs Matter) weaknesses Quality criterion mean 4.67 — judge penalized long paragraphs vs the guide-oriented rubric 1469 words slightly over the 1400 target (judge passed anyway)
2	deepseek-v4-flash	deepseek	8.1
3	mimo-v2.5	xiaomi	7.95

Beginner's Guide to Matter: Getting Started in 2026 (A/B/C arena tutorial/guide)content2026-08-19deepseek-v4-flash8.03

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	8.03	analysis strengths Highest judge mean (8.03) - PASS All required H2s present, FAQ + E-E-A-T block + HERO_IMAGE_PROMPT All 15 cited URLs verified HTTP 200, inline links only weaknesses Quality criterion mean 5.67 - long opening paragraphs vs 40-60 word target (fixed in GLM QA pass)
2	mimo-v2.5	xiaomi	7.95
3	poolside/laguna-s-2.1	openrouter	7.62

Best Smart Plug 2026: Eve vs Aqara vs Tapo (Matter/Thread vs Zigbee vs Wi-Fi)content2026-08-18poolside/laguna-s-2.18.5

#	Model	Provider	Score	Strengths / Weaknesses
1	poolside/laguna-s-2.1	openrouter	8.5	analysis strengths Best sourcing density: 8 external authority links, all HTTP 200 Exact E-E-A-T hands-off sentence present, honest scope Accurate facts: Aqara Matter via Hub M3 bridge (not native), no KP125 price, no new-for-2026 framing Complete structure: comparison table, question H2, FAQ, self-contained answers, 4 internal links
2	deepseek-v4-flash	deepseek	8.23
3	mimo-v2.5	xiaomi	7.98

Best Presence Sensor 2026: Aqara FP2 vs FP300 vs EP1 (mmWave still-presence sensors)content2026-08-18poolside/laguna-s-2.19.38

#	Model	Provider	Score	Strengths / Weaknesses
1	poolside/laguna-s-2.1	openrouter	9.38	analysis strengths Complete structure: quoted frontmatter, How We Tested, question H2, product H3s, table, verdict, FAQ 12/12 external URLs verified HTTP 200; all facts trace to official vendor pages No implied hands-on testing; honest E-E-A-T block No [N] refs, no slop words, no code fence in R3 weaknesses 1775 words exceeds the 1500 brief cap (judge passed anyway) R1 draft wrapped in a yaml code fence (fixed by R3 brief tightening)
2	deepseek-v4-flash	deepseek	8.9
3	mimo-v2.5	xiaomi	8.68

Home Assistant Green Review 2026: Still Worth It?content2026-08-17deepseek-v4-flash8.18

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	8.18	analysis strengths All facts match verified official sources Verdict-first with honest How We Tested Self-contained H2 answers with inline citations weaknesses Quality criterion lower (6.3): some H2 answers exceed 60 words
2	mimo-v2.5	xiaomi	8.17
3	poolside/laguna-s-2.1	openrouter	7.7

Aqara Doorbell Camera Hub G410 Review 2026: No-Subscription Video Doorbell (best video doorbell without subscription 2026)content2026-08-17deepseek-v4-flash6.67

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	6.67	analysis strengths All inline citations present after R3 brief fix 5 internal links with keyword-rich anchors Honest How We Tested block with canonical disclaimer 6-question FAQ, clean comparison table with dash rule weaknesses Quality criterion low (4.33): judge flagged structure/flow variance Mean below 8.0 gate (6.67)
2	poolside/laguna-s-2.1	openrouter	5.45
3	mimo-v2.5	xiaomi	0

ESPHome Starter Kit (ESK-1): First Official ESPHome Hardware at $40 — Launch News + Buying Guidecontent2026-08-14poolside/laguna-s-2.18.03

#	Model	Provider	Score	Strengths / Weaknesses
1	poolside/laguna-s-2.1	openrouter	8.03	analysis strengths Highest mean weighted score (official run 8.03; re-judge 8.12) — the only draft with reproducible >=8.0 qwen passes (3 of 4 qwen passes >= 8.05) All 12 must-include facts present with named inline citations Markdown link syntax, internal links, and FAQ citations fixed vs iteration 1 (6.93) Question-shaped H2 'Is the ESPHome Starter Kit worth it?' verbatim weaknesses Judge noise on quality criterion (3-8 across passes; 2.05 spread) — 8.03 recorded from official run 1 Pass 3 of re-judge dropped on qwen 429 (OpenRouter shared-pool quota)
2	deepseek-v4-flash	deepseek	7.8
3	mimo-v2.5	xiaomi	7.85

Petlibro Outage: Why Smart Pet Feeders Need Local Control — cloud-dependency news + local-first buyer's guidecontent2026-08-14poolside/laguna-s-2.18.03

#	Model	Provider	Score	Strengths / Weaknesses
1	poolside/laguna-s-2.1	openrouter	8.03	analysis strengths Only draft with reproducible >=8.0 qwen mean (8.03; passes 7.80/7.75/8.55) All 12 must-include facts present with named inline citations Strong sourcing (9.67) and accuracy (9.33); exact long-tail keyword H2 verbatim weaknesses Quality criterion scored 4-5 across passes (long answer paragraphs under H2s, meta description duplicated into body) — fixed in QA fill Internal-link anchors in 'Why Local Control' section were unlinked in the draft — linked in QA
2	mimo-v2.5	xiaomi	7.9
3	deepseek-v4-flash	deepseek	7.88

SHFG arena page redesign — space optimization (wide left + compact rail + real-data Process Feedback)redesign2026-08-14deepseek-v4-flash8.5

#	Model	Provider	Score
1	deepseek-v4-flash	deepseek	8.5
2	poolside/laguna-s-2.1	openrouter	7.8
3	mimo-v2.5	xiaomi	0

Matter 1.6 Update: What Changedcontent2026-08-13deepseek-v4-flash8.65

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	8.65	analysis strengths All 12 must-include facts present with correct differentiations 11 inline citations across 3+ external domains, zero numbered refs Strong hook with concrete bulb-in-box imagery Bullet-point smaller upgrades section with clear per-item payoffs weaknesses Original draft used first-person voice in E-E-A-T block (fixed in QA) Runner heuristic scored 4.5 but manual judge scored 8.65 due to citation quality
2	mimo-v2.5	xiaomi	5.5
3	poolside/laguna-s-2.1	openrouter	0

Home Assistant Automations: Getting Started in 2026content2026-08-12mimo-v2.58.57

#	Model	Provider	Score	Strengths / Weaknesses
1	mimo-v2.5	xiaomi	8.57	analysis strengths Highest mean weighted score (8.57) across 3 judge passes All 9 must-include facts present with inline citations Clean trigger-vs-condition teaching with 4 verified YAML examples FAQ properly formatted as H3 headings weaknesses Quality sub-score slightly lower (7.33) due to some H2 intros needing word-count fixes (fixed in QA)
2	deepseek-v4-flash	deepseek	8.47
3	poolside/laguna-s-2.1	openrouter	8.32

Best Smart Thermostat 2026: ecobee vs Nest vs Honeywell vs Amazoncontent2026-08-11deepseek-v4-flash8.5

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	8.5	analysis strengths Strong hook and punchy title under 65 chars All 9 verified URLs used as inline clickable citations, no numbered refs Most actionable: concrete sensor pricing, C-wire specifics, rebate advice weaknesses Initial draft had 2 factual errors (fixed in QA: Amazon display, ecobee adapter) Word count exceeded 1500 cap in raw draft (trimmed in QA) Contained seamlessly AI slop word (removed in QA)
2	mimo-v2.5	xiaomi	8.3
3	poolside/laguna-s-2.1	openrouter	0

Roborock Qrevo Curv 2 Flow Review: Best for Home Assistant?content2026-08-10deepseek-v4-flash7.33

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	7.33	analysis strengths All specs verified against Roborock product page and HA integration docs Inline citations with descriptive anchor text throughout Self-contained 40-60 word answers under every H2 Honest E-E-A-T block stating no hands-on testing weaknesses Quality criterion consistently low (5/10) across all iterations Missing HERO_IMAGE_PROMPT in final iteration (added during QA fill)
2	mimo-v2.5	xiaomi	6.78
3	poolside/laguna-s-2.1	openrouter	0

Home Assistant 2026.8 Update: What's New and What Changedcontent2026-08-07deepseek-v4-flash7.18

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	7.18	analysis strengths Best factual accuracy — all claims verified against HA release blog Complete E-E-A-T block, FAQ, self-contained H2 answers Diverse sourcing: 15 inline links to HA docs, GitHub, community threads weaknesses Judge quality scores inconsistent (4-9 range across passes)
2	mimo-v2.5	xiaomi	5.7
3	poolside/laguna-s-2.1	openrouter	4.7

ESPHome 2026.7.0: PlatformIO to ESP-IDF Default Toolchain Switchcontent2026-08-06deepseek-v4-flash9.05

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	9.05	analysis strengths E-E-A-T block links all four primary sources inline; PRs link to actual PR pages Every date, SDK version, PR number verified against primary sources; zero hallucinated claims Tight checklist-driven structure with self-contained 40-60 word answers under all H2s weaknesses Broken internal link to curtain opener (fixed post-judge) Modbus/LVGL/EN18031 compressed into single bullet; no YAML opt-out snippet
2	deepseek-v4-flash	deepseek	8.7	analysis strengths Deeper operational advice with YAML opt-out snippet and EN18031 context More detailed Modbus and Device Builder coverage weaknesses PR links point to release tag instead of actual PR pages Slightly less sourcing discipline than winner

How to Set Up Frigate NVR with Home Assistant: Getting Started in 2026content2026-08-05deepseek-v4-flash9.1

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	9.1	analysis strengths Config syntax fully valid per current Frigate docs: correct roles, detect width/height nested, correct ports shm-size formula quoted exactly and computed correctly Real inline markdown links throughout, no numbered references Clean E-E-A-T block with verification statement and honest scope disclaimer weaknesses shm-size guidance omits the docs' +40MB log allowance Storage estimate is a reasonable ballpark but not directly sourced
2	mimo-v2.5	xiaomi	7.38
3	poolside/laguna-s-2.1	openrouter	6.78

SHFG arena page redesign — widescreen density v2content2026-08-04deepseek-v4-flash9.23

#	Model	Provider	Score
1	deepseek-v4-flash	deepseek	9.23
2	mimo-v2.5	xiaomi	8.33
3	qwen3.7-flash	openrouter	6.9

Home Assistant Matter Server 9.0: The matter.js Rewritecontent2026-07-31deepseek-v4-flash9

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	9	analysis strengths Diversified inline citations from 5 URLs Clear practical tone with comprehensive migration checklist Covers all key aspects: gains, costs, and safe update steps weaknesses Could include more technical depth on GC latency impacts
2	mimo-v2.5	xiaomi	7.3
3	qwen3.7-flash	openrouter	5.3

Schlage Sense Pro Smart Deadbolt Review: UWB Hands-Free Unlock Meets Matter Over Threadcontent2026-07-27deepseek-v4-flash9

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	9
2	mimo-v2.5-pro	xiaomi	7

Zigbee 4.0 and Suzi: The Smart Home Protocol Gets a Massive Upgradecontent2026-07-24deepseek-v4-flash8.25

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	8.25
2	deepseek-v4-flash	deepseek	8

SwitchBot Lock Vision & Lock Vision Pro Review: Hands-Free Face Unlock Meets Mattercontent2026-07-23deepseek-v4-flash8.75

#	Model	Provider	Score	Strengths / Weaknesses
1	deepseek-v4-flash	deepseek	8.75
2	mimo-v2.5-pro	xiaomi	8

Home Assistant Yellow vs Raspberry Pi 5 vs Intel N100: Best HA Hardware 2026content2026-07-22mimo-v2.5-pro5.5

#	Model	Provider	Score	Strengths / Weaknesses
1	mimo-v2.5-pro	xiaomi	5.5