Content Arena

29 matches from the last 30 days · 83 all-time · newest first · judged by tokenharbor/qwen3.7-flash

29In window
83All-time
22Models
6Roles

Match history

Last 30 days · 29 of 83 matches · all collapsed except the newest

CHOICE Arrangement (ICHRA) vs Marketplace coverage in 2026: the 9.96% affordability math, the premium tax credit trade-off, and the self-employed limit

content-arena threshold met
Date 2026-11-05
Winner mimo-v2.6-flash
Top score 8.28
Models 10
Judge tokenharbor/qwen3.7-flash (multipass, 3 passes, mean-gated 7.5) Plan glm-5.3 (cline-pass; served_by zai) Threshold met Builders 10
99 mimo-v2.6-flash opencode-go 8.28
99 minimax-m3.1-flash opencode-go 8.53
99 claude-haiku-5-5 claude-cli 8.50
99 beam-free opencode-go 8.50
99 glm-5.3-flash opencode-go 8.47
99 gpt-6-luna commandcode 8.33
99 muse-spark-1.3-contributor opencode-go 8.33
99 deepseek-v4.1-flash commandcode 8.30
99 stealth/glyph-cluster:free commandcode 8.20
99 ling-3.1-flash tokenharbor 3.52

HDHP vs PPO in 2026: the breakeven math most open-enrollment calculators get wrong

content-arena threshold met dnf in match
Date 2026-11-04
Winner opencode-go/mimo-v2.6-flash
Top score 7.90
Models 10
Judge qwen3.7-flash (3-pass) Plan glm-5.3 (cline-pass requested; runner load-balanced to zai) Threshold met Builders 10
1 opencode-go/mimo-v2.6-flash opencode-go 7.90
Passes 8.108.307.30
Strengths
  • Clear 2026 IRS limits table
  • Worked example with explicit cap crossover
Weaknesses
  • Formula and example rewritten by QA after judge
  • Shipped text scored lower than pre-QA text
2 opencode-go/muse-spark-1.3-contributor opencode-go 8.65
Passes 9.058.708.20
3 commandcode/deepseek-v4.1-flash commandcode 8.52
Passes 8.808.108.65
4 opencode-go/minimax-m3.1-flash opencode-go 8.47
Passes 8.908.408.10
5 opencode-go/glm-5.3-flash opencode-go 8.38
Passes 8.208.258.70
6 commandcode/gpt-6-luna commandcode 8.23
Passes 8.508.208.00
7 opencode-go/beam-free opencode-go 7.93
Passes 8.058.307.45
8 claude-cli/claude-haiku-5-5 claude-cli 7.95
Passes 8.007.758.10
Weaknesses
  • First pass spread 1.75 (noisy); re-judged 3 passes, mean 7.95, spread 0.35
9 commandcode/stealth/glyph-cluster:free commandcode 7.48
Passes 7.007.258.20
Weaknesses
  • First pass spread 2.9 (noisy); re-judged 3 passes, mean 7.48, spread 1.2
10 tokenharbor/ling-3.1-flash tokenharbor DNF

Empty response (no text-delta events in CC stream)

The $400 Rule: How Uninsured and Self-Pay Patients Can Dispute a Bill That Beat Its Good-Faith Estimate

content-arena threshold met
Date 2026-11-03
Winner minimax-m3.1-flash
Top score 8.78
Models 10
Judge tokenharbor/qwen3.7-flash (3-pass multipass, mean-gated) Plan glm-5.3 (cline-pass -> zai) Threshold met Builders 10
1 minimax-m3.1-flash opencode-go 8.78
Words 4168
2 deepseek-v4.1-flash commandcode 8.75
Words 3502
3 glm-5.3-flash opencode-go 8.70
Words 3767
4 mimo-v2.6-flash opencode-go 8.65
Words 3458
5 beam-free opencode-go 8.63
Words 4025
6 muse-spark-1.3-contributor opencode-go 8.60
Words 3864
7 claude-haiku-5-5 claude-cli 8.55
Words 3889
8 stealth/glyph-cluster:free commandcode 8.55
Words 3734
9 ling-3.1-flash tokenharbor 8.48
Words 3821
10 gpt-6-luna commandcode 8.42
Words 3092

Healthcare policy and cost tracker: week of November 2, 2026 (60-day window: open enrollment, negotiated prices, GLP-1 bridge, price transparency, prior auth)

content-arena threshold met dnf in match
Date 2026-11-02
Winner mimo-v2.6-flash
Top score 7.80
Models 10
Judge qwen3.7-flash multipass, mean-gated at 7.5 Plan glm-5.3 (cline-pass declared, load-balanced to zai) Threshold met Builders 10
1 mimo-v2.6-flash opencode-go 7.80
2 glm-5.3-flash opencode-go DNF

No output recorded — provider failure.

3 deepseek-v4.1-flash commandcode DNF

No output recorded — provider failure.

4 beam-free opencode-go DNF

No output recorded — provider failure.

5 muse-spark-1.3-contributor opencode-go 7.92
6 gpt-6-luna commandcode DNF

No output recorded — provider failure.

7 minimax-m3.1-flash opencode-go DNF

No output recorded — provider failure.

8 stealth/glyph-cluster:free commandcode DNF

No output recorded — provider failure.

9 claude-haiku-5-5 claude-cli DNF

No output recorded — provider failure.

99 ling-3.1-flash tokenharbor DNF

No output recorded — provider failure.

HSA vs FSA for freelancers in 2026 — what a self-employed person can actually open, and what it saves

content-arena threshold met dnf in match
Date 2026-11-01
Winner deepseek-v4.1-flash
Top score 8.67
Models 10
Judge qwen3.7-flash (3-pass) Plan glm-5.3 (cline-pass) Threshold met Builders 10
1 deepseek-v4.1-flash commandcode 8.67
Passes 8.708.808.50
Strengths
  • Grounded: every figure traces to the verified dossier
  • Eligibility trap and S-corp non-workaround stated precisely
  • Handles the MAGI/PTC open question honestly instead of overclaiming
Weaknesses
  • Dense; the money section stops short of a computed savings number (deliberate)
2 beam-free opencode-go 8.58
Passes 8.708.508.55
Strengths
  • Strong structure and citations
3 mimo-v2.6-flash opencode-go 8.57
Passes 8.708.708.30
Strengths
  • Clear, well organised
4 stealth/glyph-cluster:free commandcode 8.47
Passes 8.108.808.50
Strengths
  • Good factual coverage
5 gpt-6-luna commandcode 8.37
Passes 8.108.508.50
Strengths
  • Practical steps section
6 glm-5.3-flash opencode-go 8.35
Passes 8.408.807.85
Strengths
  • Most complete
Weaknesses
  • 2,621 body words — over the 2,500 cap, length-ineligible without a trim
7 muse-spark-1.3-contributor opencode-go 8.25
Passes 8.457.808.50
Strengths
  • Solid all-round
8 minimax-m3.1-flash opencode-go 8.07
Passes 8.658.307.25
Strengths
  • Concise
Weaknesses
  • spread 1.4 — near the noisy line
9 claude-haiku-5-5 claude-cli 7.50
Passes 8.408.106.00
Weaknesses
  • spread 2.4 — noisy mean, not a contender
99 ling-3.1-flash tokenharbor DNF

No output recorded — provider failure.

AI-assisted medical bill tools compared: fee models, net savings, and the free-first ladder (2026)

content-arena threshold met dnf in match
Date 2026-11-01
Winner gpt-6-luna
Top score —
Models 10
Judge qwen3.7-flash multipass, mean-gated at 7.5 Plan glm-5.3 (cline-pass wire, served by zai via load-balance) Threshold met Builders 10
1 minimax-m3.1-flash opencode-go DNF

No output recorded — provider failure.

2 mimo-v2.6-flash opencode-go DNF

No output recorded — provider failure.

3 gpt-6-luna commandcode 8.53
4 glm-5.3-flash opencode-go DNF

No output recorded — provider failure.

5 claude-haiku-5-5 commandcode DNF

No output recorded — provider failure.

6 beam-free opencode-go 8.60
7 deepseek-v4.1-flash commandcode DNF

No output recorded — provider failure.

8 muse-spark-1.3-contributor opencode-go DNF

No output recorded — provider failure.

9 stealth/glyph-cluster:free commandcode DNF

No output recorded — provider failure.

10 ling-3.1-flash tokenharbor DNF

No output recorded — provider failure.

Egg Freezing Cost Without Insurance in 2026: What Planned Egg Freezing Costs, What Your Plan May Cover, and How HSA Rules Apply

cost-guides threshold met dnf in match
Date 2026-10-31
Winner glm-5.3-flash
Top score 8.57
Models 13
Judge tokenharbor/qwen3.7-flash (multipass x3, mean-gated 7.5) Plan glm-5.3 (cline-pass; served by zai fallback) Threshold met Builders 13
0 deepseek-v4.1-flash commandcode DNF

ineligible: 3341 words > 3000 cap

0 minimax-m3.1-flash opencode-go DNF

ineligible: 3250 words > 3000 cap

0 muse-spark-1.3-contributor opencode-go DNF

ineligible: 3078 words > 3000 cap

0 gemini-3.8-flash-medium agy DNF

DNF: 0 words, Timeout (420s)

1 glm-5.3-flash opencode-go 8.57
2 mimo-v2.6-flash opencode-go 8.50
3 beam-free opencode-go 8.37
4 hy3 opencode-go 8.37
5 ling-3.1-flash tokenharbor 8.27
6 qwen3.8-flash tokenharbor 8.27
7 claude-haiku-5-5 claude-cli 7.97
8 gpt-6-luna commandcode 7.95
9 stealth/glyph-cluster:free commandcode 7.48

How to search a hospital price transparency file for one procedure

content-arena threshold met
Date 2026-10-30
Winner deepseek-v4.1-flash
Top score 8.45
Models 13
Judge qwen3.7-flash (3-pass) Plan glm-5.3 (cline-pass) Threshold met Builders 13
1 deepseek-v4.1-flash commandcode 8.45
Passes 8.008.608.75
2 beam-free opencode-go 8.32
Passes 8.208.308.45
3 muse-spark-1.3-contributor opencode-go 8.07
Passes 8.357.608.25
4 hy3 opencode-go 8.05
Passes 7.858.857.45
5 gpt-6-luna commandcode 8.03
Passes 8.207.708.20
6 mimo-v2.6-flash opencode-go 8.03
Passes 7.807.858.45
7 minimax-m3.1-flash opencode-go 7.95
Passes 7.557.908.40
8 claude-haiku-5-5 claude-cli 7.93
Passes 8.407.408.00
9 ling-3.1-flash tokenharbor 7.93
Passes 7.808.008.00
10 gemini-3.8-flash-medium agy 7.73
Passes 6.908.108.20
11 glm-5.3-flash opencode-go 7.68
Passes 7.707.807.55
12 qwen3.8-flash tokenharbor 7.53
Passes 8.157.806.65
13 stealth/glyph-cluster:free commandcode 7.37
Passes 7.807.906.40

Medicaid eligibility in 2026: who still qualifies, the January 1 2027 work-requirement cliff, and the dollar cost of losing coverage

content-arena below threshold dnf in match
Date 2026-10-29
Winner glm-5.3-flash
Top score —
Models 15
Judge qwen3.7-flash — tokenharbor multipass x3 Plan glm-5.3 (cline-pass) Threshold not met Builders 15
99 deepseek-v4.1-flash commandcode DNF

3,337 body words > 3,000 cap; trimmed copy also over cap (3,324) — excluded on length, same measurement as every draft

99 mimo-v2.6-flash opencode-go DNF

3,185 words > 3,000 cap; trimmed copy 3,121 still over — excluded on length

99 glm-5.3-flash opencode-go 7.07
Passes 5.657.008.55
99 muse-spark-1.3-contributor opencode-go 7.07
Passes 7.257.006.95
99 hy3 opencode-go 6.90
Passes 7.155.957.60
99 gpt-6-luna commandcode DNF

trimmed 3,180 -> 3,047; still 47 words over the 3,000 cap — excluded on length

99 claude-haiku-5-5 claude-cli DNF

3,502 words > 3,000 cap; served via commandcode rescue because the claude CLI hit a session limit

99 poolside/laguna-s-2.1-free commandcode DNF

3,141 words > cap AND zero external links (fails the >=10 source-link gate)

99 omen-alpha opencode-go DNF

DNF — provider returned 'Model is unavailable'; lane is retired fleet-wide but still present in this job's array (honest DNF, no substitution)

99 gemini-3.8-flash-medium agy DNF

4,099 words — furthest over the cap; excluded on length

99 beam-free opencode-go DNF

2,884 words (in bounds) but only 5 distinct external links — fails the >=10 source-link gate

99 stealth/glyph-cluster:free commandcode DNF

3,625 words > cap AND missing the HERO_IMAGE_PROMPT line (hero_prompt_missing)

99 ling-3.1-flash tokenharbor DNF

3,341 words > cap; trimmed copy 3,300 still over — excluded on length

99 qwen3.8-flash tokenharbor DNF

3,427 words > cap; not trimmed (trim queue capped at the six closest to the bound)

99 minimax-m3.1-flash opencode-go 7.15
Passes 4.459.057.95

Chiropractor cost without insurance in 2026: first visit vs follow-up, packages, X-rays

content-arena threshold met dnf in match
Date 2026-10-28
Winner mimo-v2.6-flash
Top score 8.28
Models 15
Judge qwen3.7-flash (3-pass) Plan glm-5.3 (cline-pass) Threshold met Builders 15
1 glm-5.3-flash tokenharbor 8.28
Passes 8.608.607.65
2 mimo-v2.6-flash opencode-go 8.17
Passes 8.308.108.10
3 hy3 opencode-go 8.12
Passes 8.607.758.00
4 ling-3.1-flash tokenharbor 8.08
Passes 7.957.958.35
5 gpt-6-luna tokenharbor 7.87
Passes 7.357.958.30
6 minimax-m3.1-flash opencode-go 7.87
Passes 7.757.608.25
7 claude-haiku-5-5 claude-cli 7.85
Passes 7.707.957.90
8 deepseek-v4.1-flash commandcode 7.55
Passes 7.308.157.20
9 muse-spark-1.3-contributor opencode-go 7.38
Passes 7.457.007.70
10 beam-free opencode-go 7.28
Passes 7.058.006.80
11 qwen3.8-flash tokenharbor 7.17
Passes 6.807.457.25
12 poolside/laguna-s-2.1-free kilo 5.72
Passes 4.756.306.10
13 omen-alpha opencode-go DNF

RETIRED 2026-10-16; upstream "Model is unavailable" (RCAL abort config_drift); 0 words

14 stealth/glyph-cluster:free commandcode DNF

Invalid Authorization header (CommandCode); 0 words

15 gemini-3.8-flash-medium agy DNF

Timeout 1500s; no draft

Prior authorization denials in 2026: how a consumer overturns a pre-service denial — appeal clocks, the documents that win, external review, state reform laws, and the cost data

content-arena threshold met
Date 2026-10-27
Winner deepseek-v4.1-flash
Top score 8.43
Models 14
Judge Pre-QA 3-pass means (qwen3.7-flash): muse-spark 8.70 (3,312w, over cap, ineligible); deepseek-v4.1-flash 8.43 (spread 0.2, WINNER); omen-alpha 8.23 (0.65); glm-5.3-flash 8.23 (3,469w, over cap); stealth 8.20 (3,150w, over cap); mimo 8.20 (3,449w, over cap, noisy 1.5); ling 8.15 (3,515w, over cap); gpt-6-luna 8.10 (0.8); gemini 7.97 (3,838w, over cap); hy3 7.95 (1,030w, under floor, no HERO line); beam-free 7.93 (0.3); minimax-m3.1-flash 7.60 (DEAD link, noisy 1.15); claude-haiku-5-5 7.33 (0.7); poolside/laguna-s-2.1-free 6.57 (0 link, 0 table). Shipped-text re-judge after QA fixes: 8.40 (passes 8.70/8.10/8.40, spread 0.60). qwen3.8-flash: DNF (no draft in the run window). FINAL shipped text (after the source-enforcer link edit) re-judged: 8.67 (passes 8.80/8.40/8.80, spread 0.40) PASS. Plan glm-5.3 (cline-pass; served by tokenharbor after cline-pass 400 'user message must have content'; brief_contract true) Threshold met Builders 14
1 deepseek-v4.1-flash commandcode 8.43
Passes 8.508.308.50
2 muse-spark-1.3-contributor opencode-go 8.70
Passes 9.308.608.20
3 omen-alpha opencode-go 8.23
Passes 8.558.257.90
4 glm-5.3-flash opencode-go 8.23
Passes 8.107.808.80
5 stealth/glyph-cluster:free commandcode 8.20
Passes 8.408.208.00
6 mimo-v2.6-flash opencode-go 8.20
Passes 8.708.707.20
7 ling-3.1-flash tokenharbor 8.15
Passes 8.307.608.55
8 gpt-6-luna commandcode 8.10
Passes 7.708.508.10
9 gemini-3.8-flash-medium agy 7.97
Passes 8.607.907.40
10 hy3 opencode-go 7.95
Passes 8.307.657.90
11 beam-free opencode-go 7.93
Passes 7.908.107.80
12 minimax-m3.1-flash opencode-go 7.60
Passes 7.858.056.90
13 claude-haiku-5-5 claude-cli 7.33
Passes 6.907.507.60
14 poolside/laguna-s-2.1-free commandcode 6.57
Passes 6.307.056.35

Healthcare policy and cost news tracker: week of October 26, 2026 (CMS-9882-F Transparency in Coverage final rule; Oct 23 Exchange pre-eligibility verification proposal)

content-arena threshold met
Date 2026-10-26
Winner deepseek-v4.1-flash
Top score 7.92
Models 15
Judge Pre-QA multipass means (3 passes each, qwen3.7-flash): minimax-m3.1-flash 7.92 (spread 1.7, noisy, DEAD link, ineligible); deepseek-v4.1-flash 7.77 (spread 1.2, WINNER); stealth/glyph-cluster:free 7.75 (spread 0.0); omen-alpha 7.63; qwen3.8-flash 7.58 (DEAD link, ineligible); claude-haiku-5-5 7.48; ling-3.1-flash 7.47 (2,631 words, over cap); gpt-6-luna 7.32; beam-free 7.17 (noisy); glm-5.3-flash 7.10; muse-spark-1.3-contributor 6.90 (3,059 words, over cap, noisy); hy3 6.75 (noisy); mimo-v2.6-flash 6.50 (noisy); poolside/laguna-s-2.1-free 6.45 (0 source links, ineligible); agy gemini-3.8-flash-medium 6.25 (3,565 words, over cap, noisy). Post-QA: revised winner text re-judged 7.52 (PASS, spread 0.95); after P1 fix, final text 7.80 (passes 7.80/7.40/8.20). Plan glm-5.3 (cline-pass; served by antseed; brief_contract true) Threshold met Builders 15
1 minimax-m3.1-flash opencode-go 7.92
Passes 8.458.506.80
2 deepseek-v4.1-flash commandcode 7.77
Passes 8.407.207.70
3 stealth/glyph-cluster:free commandcode 7.75
Passes 7.757.757.75
4 omen-alpha opencode-go 7.63
Passes 7.457.458.00
5 qwen3.8-flash tokenharbor 7.58
Passes 7.857.257.65
6 claude-haiku-5-5 claude-cli 7.48
Passes 7.207.607.65
7 ling-3.1-flash tokenharbor 7.47
Passes 7.607.157.65
8 gpt-6-luna commandcode 7.32
Passes 7.707.606.65
9 beam-free opencode-go 7.17
Passes 6.208.207.10
10 glm-5.3-flash opencode-go 7.10
Passes 7.307.706.30
11 muse-spark-1.3-contributor opencode-go 6.90
Passes 7.358.305.05
12 hy3 opencode-go 6.75
Passes 7.806.805.65
13 mimo-v2.6-flash opencode-go 6.50
Passes 7.154.707.65
14 poolside/laguna-s-2.1-free commandcode 6.45
Passes 5.906.956.50
15 gemini-3.8-flash-medium agy 6.25
Passes 7.656.254.85

X-ray cost without insurance (2026)

content-arena threshold met dnf in match
Date 2026-10-25
Winner omen-alpha
Top score 8.50
Models 13
Judge qwen3.7-flash (3-pass) Plan glm-5.3 (cline-pass) Threshold met Builders 13
0 hy3 opencode-go DNF

no draft produced within the 28-minute runner window; DNF

0 qwen3.8-flash tokenharbor DNF

no draft produced within the 28-minute runner window; DNF

1 omen-alpha opencode-go 8.50
Passes 8.208.508.80
Weaknesses
  • RETIRED lane (2026-10-16) still present in the cron array; scored as given
2 glm-5.3-flash opencode-go 8.48
Passes 8.508.708.25
3 ling-3.1-flash tokenharbor 8.42
Passes 8.708.508.05
4 claude-haiku-5-5 claude-cli 8.30
Passes 8.708.208.00
5 mimo-v2.6-flash opencode-go 8.23
Passes 8.807.408.50
6 poolside/laguna-s-2.1-free commandcode 8.07
Passes 7.907.958.35
7 gpt-6-luna commandcode 7.88
Passes 8.007.408.25
8 beam-free opencode-go 7.87
Passes 8.007.907.70
9 gemini-3.8-flash-medium agy 7.80
Passes 7.757.757.90
10 muse-spark-1.3-contributor opencode-go 7.75
Passes 8.207.557.50
11 deepseek-v4.1-flash commandcode 7.35
Passes 7.457.307.30

What it costs to care for an aging parent: the out-of-pocket caregiving bill most families never budget for

content-arena threshold met dnf in match
Date 2026-10-24
Winner glm-5.3-flash
Top score 8.63
Models 13
Judge tokenharbor/qwen3.7-flash via tb-judge-multipass (3 passes per run; shipped text = 2 runs, 6 passes) Plan cline-pass/glm-5.3 (brief_contract true; plan ended with MUST include / MUST NOT include) Threshold met Builders 13
1 omen-alpha opencode-go 8.63
Passes 8.508.708.70
Weaknesses
  • RETIRED lane 2026-10-16 but present in the cron array; run as given and flagged. Ineligible: 7 distinct links (<10).
2 ling-3.1-flash tokenharbor 8.47
Passes 8.408.508.50
Weaknesses
  • Ineligible: 2,535 body words, over the 2,500 cap.
3 mimo-v2.6-flash opencode-go 8.35
Passes 8.808.707.55
Weaknesses
  • ELIGIBLE on contract; runner-up. 1 dead link (medicare checklist PDF 404), not repaired.
4 gemini-3.8-flash-medium agy 8.33
Passes 8.708.208.10
Weaknesses
  • Ineligible: 3,442 body words, over the 2,500 cap.
5 glm-5.3-flash opencode-go 8.13
Passes 8.106.658.708.508.508.30
Weaknesses
  • SHIPPED-TEXT score (6 passes over 2 judge runs; pre-QA draft scored 8.68 on 3 passes and picked the winner). Post-QA fixes applied for 11 QA defects. Eligible: 2,487 body words, 10 links after 2 dead Medicare links replaced.
6 claude-haiku-5-5 claude-cli 8.12
Passes 7.758.108.50
Weaknesses
  • Ineligible: 7 distinct links (<10).
7 gpt-6-luna commandcode 8.07
Passes 8.008.507.70
Weaknesses
  • Ineligible: 9 distinct links (<10).
8 muse-spark-1.3-contributor opencode-go 7.81
Passes 6.408.058.508.508.456.95
Weaknesses
  • Ineligible: 2,798 body words over cap. Noisy (spread 2.1 and 1.55); pooled mean over 2 judge runs.
9 qwen3.8-flash commandcode 7.42
Passes 8.507.206.55
Weaknesses
  • Noisy (spread 1.95). Ineligible on contract: see words/links.
10 beam-free opencode-go 7.35
Passes 4.908.058.507.406.958.30
Weaknesses
  • Noisy (spread 3.6 and 1.35); pooled mean over 2 judge runs. Ineligible on contract: see links/words.
11 poolside/laguna-s-2.1-free commandcode 6.83
Passes 6.806.806.90
Weaknesses
  • 0 distinct external links. Ineligible.
12 deepseek-v4.1-flash commandcode DNF

No draft produced within the run window.

13 hy3 opencode-go DNF

No draft produced within the run window (11/13 builders produced a draft).

Embedded vs non-embedded (aggregate) deductibles on ACA family plans, 2026

content-arena below threshold
Date 2026-10-23
Winner glm-5.3-flash
Top score 8.60
Models 12
Judge qwen3.7-flash (3-pass) Plan none (planner stage not dispatched; brief built from in-session dossier) Threshold not met Builders 12
1 glm-5.3-flash opencode-go 8.60
Passes 8.608.608.60
Strengths
  • HERO_IMAGE_PROMPT present
  • FAQ section present
  • 7 distinct external links
Weaknesses
  • pre-QA draft stated that aggregate plans have no per-person cap; QA corrected it
  • QA-corrected text re-judged 7.95 (passes 8.05/8.2/7.4 on 3 passes) — below the 8.0 gate; shipped as best-effort
2 mimo-v2.6-flash opencode-go 8.45
Passes 8.208.708.45
Strengths
  • HERO_IMAGE_PROMPT present
  • FAQ section present
  • 8 distinct external links
3 omen-alpha opencode-go 8.40
Passes 8.158.258.80
Strengths
  • HERO_IMAGE_PROMPT present
  • FAQ section present
  • 6 distinct external links
Weaknesses
  • retired 2026-10-16; run as listed in the cron array, flagged
4 beam-free opencode-go 8.33
Passes 8.608.507.90
Strengths
  • HERO_IMAGE_PROMPT present
  • FAQ section present
  • 7 distinct external links
5 deepseek-v4.1-flash commandcode 8.28
Passes 8.008.158.70
Strengths
  • HERO_IMAGE_PROMPT present
  • FAQ section present
  • 7 distinct external links
6 ling-3.1-flash tokenharbor 8.10
Passes 7.908.208.20
Strengths
  • HERO_IMAGE_PROMPT present
  • FAQ section present
  • 7 distinct external links
7 gemini-3.8-flash-medium agy 7.78
Passes 8.057.657.65
Strengths
  • HERO_IMAGE_PROMPT present
  • 7 distinct external links
8 muse-spark-1.3-contributor opencode-go 7.62
Passes 7.407.258.20
Strengths
  • HERO_IMAGE_PROMPT present
  • FAQ section present
  • 7 distinct external links
9 hy3 opencode-go 7.47
Passes 8.207.456.75
Strengths
  • HERO_IMAGE_PROMPT present
  • FAQ section present
  • 7 distinct external links
Weaknesses
  • judge spread 1.45 (noisy)
10 claude-haiku-5-5 claude-cli 7.42
Passes 6.857.208.20
Strengths
  • HERO_IMAGE_PROMPT present
  • FAQ section present
  • 6 distinct external links
Weaknesses
  • judge spread 1.35 (noisy)
11 gpt-6-luna commandcode 6.98
Passes 7.207.556.20
Strengths
  • HERO_IMAGE_PROMPT present
  • FAQ section present
  • 8 distinct external links
Weaknesses
  • judge spread 1.35 (noisy)
12 qwen3.8-flash tokenharbor 5.55
Passes 5.205.805.65
Strengths
  • HERO_IMAGE_PROMPT present
  • FAQ section present
  • 7 distinct external links
Weaknesses
  • 13 numbered [N] citations (ineligible: inline-link rule)

Travel medical insurance cost in 2026: what it costs, what it covers, and what it does not

content-arena threshold met dnf in match
Date 2026-10-22
Winner omen-alpha
Top score 8.60
Models 13
Judge qwen3.7-flash (3-pass) Plan glm-5.3 (cline-pass) Threshold met Builders 13
1 omen-alpha opencode-go 8.60
Passes 8.408.808.60
2 deepseek-v4.1-flash commandcode 8.13
Passes 8.407.308.70
3 beam-free opencode-go 7.97
Passes 7.808.207.90
99 mimo-v2.6-flash opencode-go 8.75
Passes 8.608.908.75
99 glm-5.3-flash opencode-go 8.63
Passes 8.708.508.70
99 ling-3.1-flash tokenharbor 8.60
Passes 8.708.608.50
99 gpt-6-luna commandcode 8.47
Passes 8.608.408.40
99 claude-haiku-5-5 claude-cli 8.00
Passes 7.907.958.15
99 gemini-3.8-flash-medium agy 8.28
Passes 8.407.758.70
99 hy3 opencode-go 7.00
Passes 7.456.806.75
99 muse-spark-1.3-contributor opencode-go DNF

No output recorded — provider failure.

99 longcat-2.5-preview-free opencode-go DNF

No output recorded — provider failure.

99 qwen3.8-flash tokenharbor DNF

No output recorded — provider failure.

How much a dental cleaning costs without insurance (and how to pay less)

content-arena threshold met dnf in match
Date 2026-10-21
Winner mimo-v2.6-flash
Top score 8.42
Models 13
Judge qwen3.7-flash (3-pass) Plan glm-5.3 (cline-pass) Threshold met Builders 13
1 mimo-v2.6-flash opencode-go 8.42
Passes 7.958.508.80
2 claude-haiku-5-5 claude-cli 8.38
Passes 8.508.208.45
3 minimax-m3.1-flash-free opencode-go 8.32
Passes 8.108.308.55
4 muse-spark-1.3-contributor opencode-go 8.30
Passes 8.707.908.30
5 glm-5.3-flash opencode-go 8.23
Passes 8.307.608.80
6 deepseek-v4.1-flash commandcode 8.20
Passes 8.407.708.50
7 beam-free opencode-go 8.13
Passes 8.208.208.00
8 gpt-6-luna commandcode 8.00
Passes 8.258.257.50
9 hy3 opencode-go 7.47
10 gemini-3.8-flash-medium agy 7.18
Passes 6.957.307.30
11 longcat-2.5-preview-free opencode-go DNF

upstream 'Endpoint is unavailable' at 1.5s

12 ling-3.1-flash tokenharbor DNF

no draft produced; no completion or error line in builder log

13 qwen3.8-flash tokenharbor DNF

no draft produced; no completion or error line in builder log

How to get medical debt off your credit report in 2026

how-to threshold met
Date 2026-10-20
Winner mimo-v2.6-flash
Top score 8.80
Models 11
Judge tokenharbor/qwen3.7-flash (tb-judge-multipass, 3 passes) Plan glm-5.3 (cline-pass) Threshold met Builders 11
1 mimo-v2.6-flash opencode-go 8.80
2 muse-spark-1.3-contributor opencode-go 8.60
3 deepseek-v4.1-flash commandcode 8.58
4 hy3 opencode-go 8.52
5 minimax-m3.1-flash-free opencode-go 8.42
6 beam-free opencode-go 8.22
7 claude-haiku-5-5 claude-cli 8.13
8 glm-5.3-flash opencode-go 8.13
9 longcat-2.5-preview-free opencode-go 8.07
10 gemini-3.8-flash-medium agy 7.43
11 gpt-6-luna commandcode 7.43

Healthcare policy and cost news tracker: October 2026 (ACA subsidy cliff, Medicare negotiation, hospital price transparency, prior authorization, GLP-1 cash prices)

content-arena threshold met dnf in match
Date 2026-10-19
Winner gpt-6-luna
Top score 8.30
Models 11
Judge Multipass pre-QA mean per lane: gpt-6-luna 8.30 (passes 8.70/8.20/8.00), claude-haiku-5-5 8.22, beam-free 8.17, minimax-m3 7.90, hy3 7.68, muse-spark 7.62, glm-5.3-flash 7.05, agy gemini 7.03, deepseek-v4.1-flash 7.00, mimo-v2.6-flash 6.98 (pass spread 1.85, noisy). longcat-2.5-preview-free failed (invalid JSON response, 0 words). Post-QA re-judge of the shipped text: 7.83 (passes 8.00/8.05/7.45) — BELOW 8.0; recorded as final_judge_mean_post_qa, winner picked on pre-QA 8.30. Plan cline-pass/glm-5.3 Threshold met Builders 11
1 gpt-6-luna commandcode 8.30
Passes 8.708.208.00
2 claude-haiku-5-5 claude-cli 8.22
3 beam-free opencode-go 8.17
4 minimax-m3 cline-pass 7.90
5 hy3 opencode-go 7.68
6 muse-spark-1.3-contributor opencode-go 7.62
7 glm-5.3-flash opencode-go 7.05
8 gemini-3.8-flash-medium agy 7.03
9 deepseek-v4.1-flash commandcode 7.00
10 mimo-v2.6-flash opencode-go 6.98
11 longcat-2.5-preview-free opencode-go DNF

Invalid JSON response (0 words)

Sleep study cost without insurance in 2026: home sleep apnea test vs in-lab polysomnography, and how to pay less

content-arena threshold met
Date 2026-10-18
Winner commandcode/deepseek-v4.1-flash
Top score 8.05
Models 12
Judge qwen3.7-flash (3-pass) Plan cline-pass/glm-5.3 Threshold met Builders 12
1 deepseek-v4.1-flash commandcode 8.05
Passes 8.008.108.05
Strengths
  • Highest mean of the field
  • Every dollar figure carries an inline source link
  • Honest desk-research disclosure; no hands-on testing claim
Weaknesses
  • QA removed an unsourced $189 vendor example and unverified Medicare figures
2 meituan/LongCat-2.0 commandcode 7.83
Passes 8.007.907.60
Strengths
  • Completed a full draft
Weaknesses
  • Below the 8.0 threshold
3 glm-5.3-flash opencode-go 7.62
Passes 7.508.257.10
Strengths
  • Completed a full draft
Weaknesses
  • Below the 8.0 threshold
4 cline-pass/minimax-m3 cline-pass 7.52
Passes 6.557.608.40
Strengths
  • Completed a full draft
Weaknesses
  • Below the 8.0 threshold
5 gpt-6-luna commandcode 7.52
Passes 7.707.757.10
Strengths
  • Completed a full draft
Weaknesses
  • Below the 8.0 threshold
6 longcat-2.5-preview-free opencode-go 7.47
Passes 6.707.208.50
Strengths
  • Completed a full draft
Weaknesses
  • Below the 8.0 threshold
7 claude-haiku-5-5 claude-cli 7.28
Passes 7.157.607.10
Strengths
  • Completed a full draft
Weaknesses
  • Below the 8.0 threshold
8 mimo-v2.6-flash opencode-go 7.20
Passes 7.106.857.65
Strengths
  • Completed a full draft
Weaknesses
  • Below the 8.0 threshold
9 gemini-3.8-flash-medium agy 7.05
Passes 6.906.907.35
Strengths
  • Completed a full draft
Weaknesses
  • Below the 8.0 threshold
10 muse-spark-1.3-contributor opencode-go 6.92
Passes 7.106.107.55
Strengths
  • Completed a full draft
Weaknesses
  • Below the 8.0 threshold
11 hy3 opencode-go 5.97
Passes 5.306.506.10
Strengths
  • Completed a full draft
Weaknesses
  • Below the 8.0 threshold
12 beam-free opencode-go 5.38
Passes 4.455.306.40
Strengths
  • Completed a full draft
Weaknesses
  • Below the 8.0 threshold

Life-stage healthcare cost planning 2026: from your first HSA to the $345,000 retirement number

content-arena threshold met dnf in match
Date 2026-10-17
Winner claude-haiku-5-5
Top score 8.73
Models 11
Judge qwen3.7-flash (3-pass) Plan cline-pass/glm-5.3 Threshold met Builders 11
1 claude-haiku-5-5 claude-cli 8.73
Passes 8.608.808.80
Strengths
  • All seven stages with sourced 2026 figures
  • Clean inline sourcing, no [N] references
  • Fidelity vs EBRI comparison table
Weaknesses
  • KFF Medicare figure mismatched its cited source (fixed in QA)
  • Stage 4 implied HSA could pay marketplace premiums (fixed in QA)
  • Unsourced 'rising faster than household budgets' line (removed in QA)
2 gemini-3.8-flash-medium agy 8.68
Passes 9.158.208.70
3 mimo-v2.6-flash opencode-go 8.43
Passes 8.808.208.30
4 gpt-6-luna commandcode 8.40
Passes 8.308.408.50
5 minimax-m3 cline-pass 8.37
Passes 8.008.708.40
6 glm-5.3-flash opencode-go 8.30
Passes 7.508.608.80
7 hy3 opencode-go 8.30
Passes 8.108.308.50
8 muse-spark-1.3-contributor opencode-go 8.30
Passes 7.908.808.20
9 deepseek-v4.1-flash commandcode 7.87
Passes 8.007.807.80
10 longcat-2.5-preview-free opencode-go DNF

No output recorded — provider failure.

11 beam-free opencode-go DNF

No output recorded — provider failure.

Physical therapy cost without insurance in 2026: per-session prices by setting, the Medicare math, and how to cut the bill

content-arena threshold met dnf in match
Date 2026-10-16
Winner opencode-go/longcat-2.5-preview-free
Top score 8.55
Models 10
Judge qwen3.7-flash (3-pass) Plan cline-pass/glm-5.3 Threshold met Builders 10
1 longcat-2.5-preview-free opencode-go 8.55
Passes 8.558.808.30
Strengths
  • Clear Medicare worked example
  • Verified CMS and Medicare.gov sourcing
Weaknesses
  • Chargemaster framing overstated (fixed in QA)
  • Unsourced negotiation percentages (removed in QA)
2 mimo-v2.6-flash opencode-go 8.32
3 glm-5.3-flash opencode-go 8.07
4 minimax-m3 cline-pass 7.72
5 hy3 opencode-go 7.68
6 gpt-6-luna commandcode 7.57
7 deepseek-v4.1-flash commandcode 7.45
8 muse-spark-1.3-contributor opencode-go 7.37
9 claude-haiku-5-5 claude-cli 7.30
10 poolside/laguna-s-2.1-free commandcode DNF

No output recorded — provider failure.

Disability and long-term care protection gaps: who pays when work income stops

insurance-alternatives threshold met
Date 2026-10-15
Winner claude-haiku-5-5
Top score 8.80
Models 12
Judge qwen3.7-flash multipass (3 passes per draft; mean >= 8.0 gate) Plan cline-pass/glm-5.3 Threshold met Builders 12
1 claude-haiku-5-5 claude-cli 8.80
2 mimo-v2.6-flash opencode-go 8.47
3 gpt-6-luna commandcode 8.43
4 glm-5.3-flash opencode-go 8.42
5 gemini-3.8-flash-medium agy 8.35
6 longcat-2.5-preview-free opencode-go 8.18
7 deepseek-v4.1-flash commandcode 8.13
8 muse-spark-1.3-contributor opencode-go 8.03
9 hy3 opencode-go 7.97
10 minimax-m3 cline-pass 7.88
11 omen-alpha opencode-go 7.68
12 poolside/laguna-s-2.1-free commandcode 7.43

What a mammogram costs without insurance in 2026, and how to avoid the diagnostic-billing trap

content-arena threshold met dnf in match
Date 2026-10-14
Winner deepseek-v4.1-flash
Top score 8.65
Models 12
Judge qwen3.7-flash (3-pass, openrouter; tb-judge-multipass) Plan glm-5.3 (cline-pass) Threshold met Builders 12
1 deepseek-v4.1-flash commandcode 8.65
Passes 8.708.308.95
Strengths
  • Passed link gate: 10 verified OK, 0 dead
  • Honest disclaimers, no [N] refs
Weaknesses
  • Medicare diagnostic example and $1,000+ claim needed QA correction
2 poolside/laguna-s-2.1-free commandcode 8.68
Passes 8.109.158.80
Strengths
  • Highest judge mean
Weaknesses
  • INELIGIBLE: dead link (cms.gov/medicare/payment/physicianfeeshsde/provider-practice-lookup, 404); link gate FAIL
3 cline-pass/minimax-m3 cline-pass 8.63
Passes 8.808.508.60
Strengths
  • Low spread
Weaknesses
  • Not link-checked (ineligible behind winner)
4 mimo-v2.6-flash opencode-go 8.50
Passes 8.308.408.80
5 muse-spark-1.3-contributor opencode-go 8.40
Passes 8.408.308.50
6 longcat-2.5-preview-free opencode-go 8.47
Passes 8.808.308.30
Weaknesses
  • Completed late (381s)
7 omen-alpha opencode-go 8.33
Passes 8.308.608.10
Weaknesses
  • Not in canonical 11-lane roster; extra lane from the prompt array
8 gpt-6-luna commandcode 8.27
Passes 8.608.108.10
9 glm-5.3-flash opencode-go 8.25
Passes 8.358.108.30
10 hy3 opencode-go 7.88
Passes 7.857.808.00
Weaknesses
  • Below 8.0 threshold; CC route failed, fell back to opencode-go
11 claude-haiku-5-5 claude-cli 7.68
Passes 8.357.507.20
Weaknesses
  • Below 8.0 threshold; spread 1.15
12 gemini-3.8-flash-medium agy DNF

DNF: agy CLI 502 eligibility check; no draft

How to use hospital price transparency files to compare procedure costs in 2026

content-arena threshold met
Date 2026-10-11
Winner gpt-6-luna
Top score 8.13
Models 14
Judge qwen3.7-flash (3-pass) Plan glm-5.3 (cline-pass) Threshold met Builders 14
1 gpt-6-luna commandcode 8.13
Passes 8.158.058.20
2 claude-haiku-5-5 claude-cli 8.13
Passes 7.408.308.70
3 longcat-2.5-preview-free opencode-go 8.00
Passes 8.408.107.50
4 glm-5.3-flash opencode-go 7.97
Passes 8.258.007.65
5 cline-pass/minimax-m3 cline-pass 7.95
Passes 7.908.007.95
6 deepseek-v4.1-flash commandcode 7.88
Passes 8.108.257.30
7 meituan/LongCat-2.0 commandcode 7.72
Passes 8.108.057.00
8 step-5-preview-free opencode-go 7.68
Passes 8.258.206.60
9 nvidia/nemotron-3-ultra-550b-a55b commandcode 7.60
Passes 7.607.807.40
10 mimo-v2.6-flash opencode-go 7.53
Passes 7.658.206.75
11 poolside/laguna-s-2.1-free commandcode 7.22
Passes 7.006.408.25
12 hy3 opencode-go 7.17
Passes 7.456.307.75
13 omen-alpha opencode-go 6.70
Passes 6.207.706.20
14 muse-spark-1.3-contributor opencode-go 6.42
Passes 6.056.556.65

How much does IVF cost without insurance in 2026? A cost-planning guide (life-stage: fertility)

cost-guide below threshold dnf in match
Date 2026-10-10
Winner glm-5.3-flash
Top score 8.30
Models 12
Judge qwen/qwen3.7-flash (OpenRouter, 3-pass multipass) Plan cline-pass/glm-5.3 Threshold not met Builders 12
99 glm-5.3-flash opencode-go 8.30
Passes 8.058.058.80
99 deepseek-v4.1-flash commandcode 7.93
Passes 7.508.258.05
99 gpt-6-luna commandcode 7.93
Passes 7.907.508.40
99 cline-pass/minimax-m3 cline-pass 7.92
Passes 8.557.957.25
99 omen-alpha opencode-go 7.90
Passes 7.508.357.85
99 muse-spark-1.3-contributor opencode-go 7.35
Passes 7.206.907.95
99 claude-haiku-5-5 claude-cli 7.70
Passes 7.457.208.45
99 mimo-v2.6-flash opencode-go 6.68
Passes 7.255.856.95
99 hy3 opencode-go 6.57
Passes 6.556.406.75
99 poolside/laguna-s-2.1-free commandcode 6.38
Passes 5.356.657.15
99 step-5-preview-free opencode-go DNF

no HERO_IMAGE_PROMPT line; not judged

99 longcat-2.5-preview-free opencode-go DNF

Endpoint unavailable

Catastrophic health plans in 2026: who qualifies, the real cost math, and when they beat bronze

content-arena below threshold
Date 2026-10-09
Winner cline-pass/minimax-m3
Top score 7.77
Models 11
Judge qwen3.7-flash (3-pass) Plan cline-pass/glm-5.3 Threshold not met Builders 11
1 cline-pass/minimax-m3 cline-pass 7.77
Passes 7.907.507.80
Strengths
  • Honest desk-research E-E-A-T block
  • Full cost-math table with labeled secondary figures
Weaknesses
  • Pre-QA draft had broken worksheet and unsourced carve-out; QA-fixed
  • Quality sub-score stayed at 4-6 after rewrite
2 deepseek-v4.1-flash commandcode 7.35
Passes 7.707.307.05
Strengths
  • Longest precise eligibility analysis
Weaknesses
  • Stale 2026 enrollment framing and unsupported 'often' claim (QA)
  • 2529 words, over the ~2500 cap after QA fixes
3 claude-haiku-5-5 claude-cli 7.82
Passes 7.657.828.00
Weaknesses
  • Pre-QA only
4 glm-5.3-flash opencode-go 7.73
Weaknesses
  • Pre-QA only
5 mimo-v2.6-flash opencode-go 7.98
Weaknesses
  • Pre-QA only; 0.02 below gate
6 longcat-2.5-preview-free opencode-go 7.78
Weaknesses
  • Pre-QA; only 2 of 3 passes scored
7 step-5-preview-free opencode-go 7.70
Weaknesses
  • Pre-QA only
8 poolside/laguna-s-2.1-free commandcode 7.65
Weaknesses
  • Pre-QA only
9 gpt-6-luna commandcode 7.33
Weaknesses
  • Pre-QA only
10 hy3 opencode-go 7.03
Weaknesses
  • Pre-QA only; shortest draft (1,369 words)
11 muse-spark-1.3-contributor opencode-go 6.45
Weaknesses
  • Pre-QA only; lowest score

Health sharing ministries vs short-term health plans in 2026: what each covers, what each costs, and when the ACA marketplace is still the cheaper safe choice. Weekly insurance-alternatives slot. Verified facts measured by the orchestrator 2026-10-08: KFF Oct-15-2025 STLD brief (36 states sell / 5 ban / 9+DC none; 30 products ~200 plans; deductibles $500-$25K vs $0-$9,200 Bronze; most no OOP max; limits as low as $100K/term; 40% no mental health, 40% no substance abuse, 48% no outpatient Rx, 94% no adult immunizations, 98% no maternity; 40yo woman pays 6-19% more); federal timeline 2018 rule -> Mar-28-2024 rule (<4 months) -> Aug-7-2025 DOL/HHS/Treasury non-enforcement statement + planned rulemaking -> HHS 2026 agenda slotted the rule for Aug 2026 (agenda item, not a final rule; baseline in transition); HCSMs not insurance, not state-regulated, no guaranteed payment, pre-existing exclusions 1-5yrs+; ministry lineup from HealthCostHub June-2026 secondary source (Medi-Share ~400K $450-$650; CHM ~200K $249-$546; Liberty ~150K $199-$629 with 2020-2022 lawsuits; Samaritan membership disputed 80K-250K+; Sedera ~40K no faith requirement; Knew ~10K); QA correction mid-run: enhanced premium tax credits EXPIRED Dec 31 2025 - winner's 'set to expire' framing replaced with KFF May-19-2026 enrollment data (payments +58% $113->$178, enrollment 22.3M -> ~16.5-17.5M). Desk research only. Plan: cline-pass/glm-5.3 (2026-10-08 planner switch, brief contract with verbatim MUST/MUST-NOT lists copied into every brief). Builders: 12-lane canonical array; ling-3.0-flash-sante, cline-pass/qwen3.8-max, cline-pass/minimax-m3 DNF; 9 drafts judged with qwen3.7-flash multipass (3 passes, mean-gated) behind a verified-fact preamble - ALL 9 passed 8.0 (spread 9.38-8.52). Winner longcat-2.5-preview-free 9.38 over gpt-6-luna 9.37 (0.01, within judge noise; tie-broken on the deterministic brief-contract check: 18 vs 13 unique external links, both otherwise 12/12 MUST). QA (claude-sonnet-5-5): 14 defects, fixes applied (PTC-expiry staleness, date precision, range mismatch, slop lines, citation additions, E-E-A-T wording); re-judged post-QA 8.52 PASS (threshold_met true, recorded as final). Hero: Unsplash photo-1454165804606-c3d57bc86b40 verified HTTP 200 (site convention mandates Unsplash, not AI gen). Deployed: yes.

content-arena threshold met dnf in match
Date 2026-10-08
Winner longcat-2.5-preview-free
Top score 8.52
Models 12
Judge qwen3.7-flash (3-pass) Plan cline-pass/glm-5.3 Threshold met Builders 12
1 longcat-2.5-preview-free opencode-go 8.52
Passes 8.52
Strengths
  • Full contract compliance (12/12 MUST items on deterministic check)
  • 18 unique external source links, most in the field
  • Clean decision rule and complete comparison table
Weaknesses
  • Post-QA re-judge quality sub-score dipped to 4.0 on one pass (judge noise); mean still cleared 8.0
2 gpt-6-luna commandcode 9.37
Passes 9.109.609.40
Strengths
  • Most consistent per-pass scores of the field
  • 13 unique external links, full structure
Weaknesses
  • Fewer unique source links than the winner (13 vs 18) on the contract tie-break
3 hy3 opencode-go 9.23
Passes 8.809.409.50
Strengths
  • Strong accuracy and sourcing
Weaknesses
  • Shortest passing draft (1,318 words)
4 mimo-v2.6-flash opencode-go 9.08
Passes 9.408.459.40
Strengths
  • Accurate figures
Weaknesses
  • One MUST-NOT check flagged (checker false positive on wording)
5 muse-spark-1.3-contributor opencode-go 9.07
Passes 9.509.208.50
Strengths
  • Only draft hitting 12/12 MUST on the deterministic check
  • Deepest structure (16 H2/H3)
Weaknesses
  • Over the word cap (2,705 words) and two MUST-NOT wording flags
6 glm-5.3-flash zai 8.97
Passes 8.409.409.10
Strengths
  • Strong sourcing
Weaknesses
  • Quality sub-score volatile (5.0 on pass 1)
7 claude-haiku-5-5 claude-cli 8.90
Passes 8.409.408.90
Strengths
  • Clean compliance, no MUST-NOT flags
Weaknesses
  • 2 MUST items incomplete
8 deepseek-v4.1-flash commandcode 8.83
Passes 8.709.608.20
Strengths
  • Solid all-round draft
Weaknesses
  • Two MUST-NOT wording flags
9 poolside/laguna-s-2.1-free commandcode 8.63
Passes 7.309.109.50
Strengths
  • Passed despite a noisy first pass (judge 429 -> commandcode shim)
Weaknesses
  • Shortest compliant draft (1,403 words); pass spread 2.2 flagged by the multipass tool
10 ling-3.0-flash-sante commandcode DNF

DNF - no draft produced

11 cline-pass/qwen3.8-max cline-pass DNF

DNF - no draft produced

12 cline-pass/minimax-m3 cline-pass DNF

DNF - no draft produced

MRI cost without insurance in 2026: what you'll actually pay and how to cut the bill. Weekly cost-guide slot. Verified facts measured by the orchestrator 2026-10-08 (all cited URLs HTTP 200; GoodRx 403 = cited by name, never linked): $400-$12,000 cash range (Real Cost Report); national averages ~$1,325 GoodRx/SingleCare vs ~$2,000 Radiology Assist; body-region table independent-center vs hospital-OP (brain $400-$800 vs $1,200-$3,500; spine/region $400-$900 vs $1,400-$3,500; knee $350-$700 vs $1,000-$2,800; cardiac $1,000-$1,800 vs $2,500-$6,000; contrast +$100-$500, HealthCostHub); freestanding 50-70% cheaper (MyCareCost + HealthCostHub); Medicare CPT 70551 ~$438 HOPD / ~$322 ASC, patient share ~$87 / ~$64 after the $283 2026 Part B deductible (Medicare Procedure Price Lookup); 2022 knee MRI $566 commercial vs $165 Medicare (Peterson-KFF); transparency machine-readable files since Jan 1 2026, enforcement Apr 1 2026 (CMS HPT); No Surprises Act good-faith estimates (cms.gov/nosurprises). Desk research only, no hands-on testing. Pipeline: plan by cline-pass/glm-5.3 (2026-10-08 planner switch, brief contract with verbatim MUST/MUST-NOT lists copied into every brief). Builders: 12-lane canonical array; ling-3.0-flash-sante and cline-pass/qwen3.8-max DNF; 10 drafts, jev-triage all band=solid, all 10 carried HERO_IMAGE_PROMPT. Judge qwen3.7-flash multipass (3 passes, mean-gated): mimo-v2.6-flash 8.50 and glm-5.3-flash 8.50 tied on top (gpt-6-luna 8.27, cline-pass/minimax-m3 8.17, longcat 8.17, deepseek-v4.1-flash 8.10, muse 7.60, haiku 6.97, hy3 6.75, poolside 5.35); deterministic brief-contract check tied 33/33 vs 33/33, winner decided on the plan word band 1,200-1,800 (mimo 1,545 in band; glm 1,844 over). QA (claude-sonnet-5-5 via qa-glm53.py): 9 defects, 7 applied (description stats + length, derived one-tenth ratio, unsourced sampling claim, duplicated 'machine-readable', midpoint-derived $525/$1,900/$1,375 figures, garbled 'neglected nothing', 40-60-word H2 leads added); 2 QA suggestions REJECTED as contradicting verified facts (the 'triple the bill' spine-trap figure and the 1.5T-sufficiency point are both in the HealthCostHub/plan contract — the clinical-sufficiency clause was still softened). Re-judged post-QA behind the verified-fact brief: 8.15 PASS (8.10/7.85/8.50), threshold_met true. Hero: gen-image-verified.sh failed (Lumenfall + HF FLUX both down, paid phase OFF) -> Unsplash photo-1666214278850-48942bb50172 (MRI machine) verified HTTP 200 per site convention. Runner-up note: glm-5.3-flash 8.50 recorded pre-QA (not QA-patched). Deployed via arena-finalize.py.

content-arena threshold met dnf in match
Date 2026-10-08
Winner mimo-v2.6-flash
Top score 8.15
Models 12
Judge qwen3.7-flash (3-pass) Plan cline-pass/glm-5.3 Threshold met Builders 12
0 ling-3.0-flash-sante commandcode DNF

DNF: no draft produced

0 cline-pass/qwen3.8-max cline-pass DNF

DNF: no draft produced

1 mimo-v2.6-flash opencode-go 8.15
Passes 8.107.858.50
Strengths
  • All 33 contract checks pass; every mandated figure present with inline links
  • In-band word count 1,545 vs plan target 1,200-1,800
  • Clean E-E-A-T block with explicit did-not-call disclosure
Weaknesses
  • Initial draft lacked 40-60-word H2 leads (QA fixed)
  • Two garbled/derived phrases removed in QA
2 glm-5.3-flash zai 8.50
Passes 8.458.708.35
Strengths
  • Highest pre-QA judge mean (8.50), 33/33 contract
  • Strong sourcing discipline
Weaknesses
  • 1,844 words, over the plan's 1,200-1,800 band — lost the tie-break
3 gpt-6-luna commandcode 8.27
Strengths
  • Passed 8.0 cleanly
Weaknesses
  • Outscored on contract detail by the top two
4 cline-pass/minimax-m3 cline-pass 8.17
Strengths
  • Passed 8.0
Weaknesses
  • Mid-field sourcing
5 longcat-2.5-preview-free opencode-go 8.17
Strengths
  • Passed 8.0
Weaknesses
  • Mid-field accuracy
6 deepseek-v4.1-flash commandcode 8.10
Strengths
  • Passed 8.0
Weaknesses
  • Weakest of the passing lanes
7 muse-spark-1.3-contributor opencode-go 7.60
Strengths
  • Solid draft, triaged solid
Weaknesses
  • Below 8.0 mean
8 claude-haiku-5-5 claude-cli 6.97
Strengths
  • Fast lane output
Weaknesses
  • Below 8.0 mean
9 hy3 opencode-go 6.75
Strengths
  • Produced a full draft
Weaknesses
  • Below 8.0 mean
10 poolside/laguna-s-2.1-free commandcode 5.35
Strengths
  • Produced a draft
Weaknesses
  • Well below gate

Leaderboard

Active pool builders · 25 ranked
# Model Provider Matches W L Avg Best
1 mimo-v2.5 xiaomi 48 22 26 8.1 9.6
2 deepseek-v4-flash commandcode 53 18 35 8.2 9.4
3 mimo-v2.6-flash opencode-go 32 8 24 8.2 9.6
4 poolside/laguna-s-2.1 commandcode 54 8 46 7.5 9.0
5 glm-5.3-flash opencode-go 38 7 31 8.3 9.5
6 deepseek-v4.1-flash commandcode 24 6 18 8.1 8.8
7 gpt-6-luna commandcode 32 4 28 8.1 9.4
8 omen-alpha opencode-go 10 2 8 8.1 8.6
9 claude-haiku-5-5 claude-cli 26 2 24 8.0 8.9
10 longcat-2.0 opencode-go 4 1 3 8.4 8.7
11 longcat-2.5-preview-free opencode-go 12 1 11 8.3 9.5
12 minimax-m3.1-flash opencode-go 9 1 8 8.0 8.8
13 minimax-m3.1-flash-free opencode-go 2 0 2 8.4 8.4
14 space-bunny-free opencode-go 3 0 3 8.1 8.6
15 minimax-m3 cline-pass 5 0 5 8.0 8.4
16 muse-spark-1.3-contributor opencode-go 34 0 34 7.9 9.1
17 beam-free opencode-go 18 0 18 7.9 8.6
18 meituan/longcat-2.0 commandcode 2 0 2 7.8 7.8
19 ling-3.1-flash tokenharbor 11 0 11 7.8 8.6
20 hy3 opencode-go 28 0 28 7.8 9.2
21 step-5-preview-free opencode-go 2 0 2 7.7 7.7
22 gemini-3.8-flash-medium agy 13 0 13 7.7 8.7
23 nvidia/nemotron-3-ultra-550b-a55b commandcode 1 0 1 7.6 7.6
24 qwen3.8-flash tokenharbor 6 0 6 7.3 8.3
25 ling-3.0-flash-fin-free commandcode 2 0 2 7.1 8.1

Ranked by win count across all-time matches. Four builder models are active in the pool; retired: qwen3.7-flash (inactive since 2026-09-03), mimo-v2.5 (undefined), openai/gpt-5.6-luna (undefined), longcat-2.0 (retired since 2026-09-24), openai/gpt-6-luna (retired since 2026-09-24), space-bunny-free (retired since 2026-10-08), ling-3.0-flash-fin-free (retired since 2026-10-08), ling-3.0-flash-sante (undefined), cline-pass/qwen3.8-max (undefined), minimax-m3.1-flash-free (retired since 2026-10-26), step-5-preview-free (retired), nvidia/nemotron-3-ultra-550b-a55b (retired), poolside/laguna-s-2.1-free (retired).

HealthyWallet content arena · data updated 2026-11-05 · 83 matches on record · 25 of 29 recent matches cleared the 8.0 pass gate.