https://2026cnmllmtierlist.datacorp.top/en/articles/ Rankings, methodology, vertical field guides and the cold-start log for 32 China-native multimodal LLMs. Every essay is bound to an artefact recomputable only on this site. en CC BY 4.0 · 2026 CNMM Tier List cnmm-tierlist https://www.rssboard.org/rss-specification Sun, 18 Oct 2026 08:00:00 +0800 https://2026cnmllmtierlist.datacorp.top/en/articles/anti-sycophancy-index/ https://2026cnmllmtierlist.datacorp.top/en/articles/anti-sycophancy-index/ Sun, 18 Oct 2026 08:00:00 +0800 Every benchmark tests whether a model is right the first time; none tests how many rounds it takes to cave when you insist on a wrong answer. We build an anti-sycophancy index for all 32 models: 41.7% average capitulation, 1.7 rounds to cave, expert pressure at 49% and consistency pressure at 55% against 34% for plain insistence — and visual questions run about 12 points above knowledge questions. en https://2026cnmllmtierlist.datacorp.top/en/articles/model-self-knowledge/ https://2026cnmllmtierlist.datacorp.top/en/articles/model-self-knowledge/ Sun, 18 Oct 2026 08:00:00 +0800 We asked all 32 models the same six spec questions about themselves and checked each answer against our dossiers. Vendor scores 93.8% correct, parameter count only 15.6%, for an average of 2.9 of 6; 22 models claim to be their family’s latest; and self-knowledge versus composite is r = 0.09, essentially zero. en https://2026cnmllmtierlist.datacorp.top/en/articles/context-window-economics/ https://2026cnmllmtierlist.datacorp.top/en/articles/context-window-economics/ Sun, 18 Oct 2026 08:00:00 +0800 Windows across 32 models run from 32K to 2M with a 128K median, and only 12 reach 256K or more. Window versus price is r = 0.41, R² = 0.17 — not a pricing anchor. More important: nominal window is not effective depth. The 12 models at ≥256K run 38%–68% efficient, and 2M Kimi K3 has an effective depth of 780K. en https://2026cnmllmtierlist.datacorp.top/en/articles/chinese-dialect-field-guide/ https://2026cnmllmtierlist.datacorp.top/en/articles/chinese-dialect-field-guide/ Sun, 18 Oct 2026 08:00:00 +0800 Public benchmarks test standard Mandarin only. We build a mini tier list over eight dialect and accent classes — Cantonese, Wu, Hokkien, Sichuanese, Northeastern, Central Plains, minority languages and code-switching. Only 11 of 32 models have an audio modality, and on the weighted dialect board Spark X2 climbs from 18th to 1st while WITA-Omni rises from 23rd to 5th. en https://2026cnmllmtierlist.datacorp.top/en/articles/model-naming-archaeology/ https://2026cnmllmtierlist.datacorp.top/en/articles/model-naming-archaeology/ Sun, 18 Oct 2026 08:00:00 +0800 We split all 32 model names into tokens — brand, generation, size (Max/Pro), speed (Flash), modality (Omni/VL/Vision) and pure codenames. Grouped into five families by the most prominent word, a name discriminates price by about 17× (¥15.0 vs ¥0.9) and capability by only about 1.07× (93.5 vs 87.7); the five carrying Omni/VL/Vision average 84.1, below the site-wide 85.0. en https://2026cnmllmtierlist.datacorp.top/en/articles/model-cross-review/ https://2026cnmllmtierlist.datacorp.top/en/articles/model-cross-review/ Sun, 18 Oct 2026 08:00:00 +0800 What happens when models score models? This site ran a reproducible cross-review: eight models as judges, blind-scoring the same outputs from all 32, measuring three biases — self-preference (rating one’s own family higher), family bias (same-vendor mutual inflation) and score compression (strong judges flatten differences). The finding: the bias is structural, and both its direction and magnitude are predictable. This is the first public Chinese multimodal cross-review matrix anywhere. en https://2026cnmllmtierlist.datacorp.top/en/articles/industry-exclusivity-map/ https://2026cnmllmtierlist.datacorp.top/en/articles/industry-exclusivity-map/ Sun, 18 Oct 2026 08:00:00 +0800 Comparisons ask “who is stronger”; procurement asks “who can do my scenario”. This 2026 China native multimodal LLM tier list covers 32 models, at least 15 of which appear in no mainstream comparison — spread across 15 vertical scenarios with only one or two contenders each. This essay re-sorts the whole board by scenario rather than by model: finance, e-commerce, games, security, medical, industrial, government, education, content, customer service, open source, edge, embodied, research and recognition — a “scenario to only answer” map. en https://2026cnmllmtierlist.datacorp.top/en/articles/security-video-field-guide/ https://2026cnmllmtierlist.datacorp.top/en/articles/security-video-field-guide/ Sun, 18 Oct 2026 08:00:00 +0800 Public video benchmarks test film clips, variety shows and YouTube highlights — not one tests surveillance footage. This 2026 China native multimodal LLM tier list contains exactly one model trained for security: Hikvision Guanlan 3. This essay builds a separate mini tier list over eight surveillance classes — night-vision IR, perimeter intrusion, cross-camera tracking, crowd density, high-altitude projectiles, fire and smoke, plate recognition and safety compliance — and prices surveillance video (minute-level recall) separately from general video (second-level timing). en https://2026cnmllmtierlist.datacorp.top/en/articles/gefei-method-review/ https://2026cnmllmtierlist.datacorp.top/en/articles/gefei-method-review/ Sun, 18 Oct 2026 08:00:00 +0800 “Find a low-competition term, ship fast, get indexed in a week, rank #1” is repeated everywhere. We split it into five falsifiable hypotheses and audit each, then answer the question the retellings avoid: can AI-generated content actually get indexed — and why human-written filler ends up “crawled – currently not indexed” instead. en https://2026cnmllmtierlist.datacorp.top/en/articles/cold-models-survival/ https://2026cnmllmtierlist.datacorp.top/en/articles/cold-models-survival/ Sun, 18 Oct 2026 08:00:00 +0800 Public comparisons test the same dozen flagships. At least nine of our 26 appear in no mainstream Chinese comparison: SenseTime, StepFun, NeuroXess, InternVL, Kuaishou, Meituan, AgiBot, Huawei and ModelBest. This essay does not rehabilitate them — it answers one question: when a model holds no popular slot on any board, what earns it a test and a purchase order. en https://2026cnmllmtierlist.datacorp.top/en/articles/price-gaps/ https://2026cnmllmtierlist.datacorp.top/en/articles/price-gaps/ Sun, 18 Oct 2026 08:00:00 +0800 Twenty-six priced models use only 20 price points; the median is ¥6 and only eight sit on the price-performance frontier. From ¥0.6 to ¥2.5 costs 4.2× for 1.3 points; from ¥15 to ¥18 costs ¥3 for 0.2. Two bands are empty: ¥2.5–4 and ¥12–15. Price and score correlate at 0.44, explaining 19%. en https://2026cnmllmtierlist.datacorp.top/en/articles/one-model-teams/ https://2026cnmllmtierlist.datacorp.top/en/articles/one-model-teams/ Sun, 18 Oct 2026 08:00:00 +0800 Thirty-two models come from 27 teams: Alibaba 4, Zhipu 2, DeepSeek 2 and 24 teams with one each. Single point is not small company — ByteDance, Tencent, Baidu, Xiaomi and Huawei all field one; it is not weak either — half the top ten comes from single-point teams. The structure is three product lines plus twenty-four single points. en https://2026cnmllmtierlist.datacorp.top/en/articles/model-age-vs-rank/ https://2026cnmllmtierlist.datacorp.top/en/articles/model-age-vs-rank/ Sun, 18 Oct 2026 08:00:00 +0800 Release dates against composite across 32 models: ages from 9 to 418 days, Pearson r -0.11, R² 0.012; every 100 days costs 0.6 points. But age correlates with evidence (r = 0.438): grade A averages 213 days, B 125, C 86. Time cannot buy score, but it buys evidence. en https://2026cnmllmtierlist.datacorp.top/en/articles/zero-price-models/ https://2026cnmllmtierlist.datacorp.top/en/articles/zero-price-models/ Sun, 18 Oct 2026 08:00:00 +0800 Six of 32 models price output at ¥0: WITA, Emu3.5, Zidong Taichu 2.0, Guanlan 3, NeuroXess-N1 and BAGEL-2. None is OSI open source; none is download-and-commercialise. We split zero price into three types and TCO into five lines: licence, compute, integration, compliance and exit. en https://2026cnmllmtierlist.datacorp.top/en/articles/modality-negative-space/ https://2026cnmllmtierlist.datacorp.top/en/articles/modality-negative-space/ Sun, 18 Oct 2026 08:00:00 +0800 Thirty-two models reduce to eight modality bundles: text plus image is the floor, video covers 24, audio 11, GUI 4, embodied 2. No model is 6/6; none combines embodied and GUI. We map the negative space and separate two real gaps from two pseudo-gaps. en https://2026cnmllmtierlist.datacorp.top/en/articles/opacity-index/ https://2026cnmllmtierlist.datacorp.top/en/articles/opacity-index/ Sun, 18 Oct 2026 08:00:00 +0800 Only five of 32 models disclose a parameter count; 24 give an adjective, three say nothing. Context disclosure is 96.9%, price 81.3%. We put a transparent discount on the unknown: evidence A/B/C costs 0/1.5/3.5 points, plus 0.5/0.5/1.0 for opaque parameters, price and context — and the top spot moves from Qwen3.8-Max to Doubao. en https://2026cnmllmtierlist.datacorp.top/en/articles/model-shapes/ https://2026cnmllmtierlist.datacorp.top/en/articles/model-shapes/ Sun, 18 Oct 2026 08:00:00 +0800 A composite collapses eight dimensions into one number; procurement buys the vector. We sort all 32 models by spread into plain, slope, valley and cliff: the top ten average 10.3, the bottom ten 25.7. The overall board rewards “no hole”; single-dimension boards reward “the longest plank”. en https://2026cnmllmtierlist.datacorp.top/en/articles/dimension-redundancy/ https://2026cnmllmtierlist.datacorp.top/en/articles/dimension-redundancy/ Sun, 18 Oct 2026 08:00:00 +0800 We correlated all eight scores across 32 models. Document and chart move at r = 0.94; agent and ecosystem at 0.79. The first principal component explains 62.9% of variance and the participation ratio puts the scorecard at 2.33 effective dimensions. The dimensions are not useless; the weighted sum is voting twice for the same thing. en https://2026cnmllmtierlist.datacorp.top/en/articles/frontier-wave-2026q4/ https://2026cnmllmtierlist.datacorp.top/en/articles/frontier-wave-2026q4/ Sat, 17 Oct 2026 08:00:00 +0800 Between 30 September and 9 October 2026, four frontier models shipped within eleven days. Instead of paraphrasing four launch posts, we put them on one alignment table: who raised prices, who cut them, who moved capability from generation to agents, who turned a specialist into a generalist — and what a serious buyer should do when all four arrive at evidence grade B. en https://2026cnmllmtierlist.datacorp.top/en/articles/cold-start-experiment/ https://2026cnmllmtierlist.datacorp.top/en/articles/cold-start-experiment/ Sat, 17 Oct 2026 08:00:00 +0800 Google decides something is “worth indexing” by whether it has unique value, not by whether a human wrote it. This is a publicly checkable experiment: no backlinks, no paid traffic — one long-tail term, 126 pages and structured data, watching indexing speed and rank movement, publishing every result including the failures. en https://2026cnmllmtierlist.datacorp.top/en/articles/enterprise-selection-playbook/ https://2026cnmllmtierlist.datacorp.top/en/articles/enterprise-selection-playbook/ Fri, 16 Oct 2026 08:00:00 +0800 A capability tier list is not a procurement tier list. We apply five gates in sequence — commercial licence, private deployment, an enterprise capability floor, official domestic-accelerator support, and evidence grade A — and count how many of the 26 each gate removes. One model survives all five, and it is neither #1 nor the strongest of the survivors. en https://2026cnmllmtierlist.datacorp.top/en/articles/cost-per-1000-images/ https://2026cnmllmtierlist.datacorp.top/en/articles/cost-per-1000-images/ Thu, 15 Oct 2026 08:00:00 +0800 Qwen3.8-Flash and DeepSeek-V4.1-Flash dragged the price line down to ¥0.6–1.2 and redrew this tier list’s value frontier. Using the strict definition of Pareto dominance, we split the 26 models into “on the frontier” and “strictly dominated” — the latter means another model is cheaper and no worse on any dimension. en https://2026cnmllmtierlist.datacorp.top/en/articles/gui-agent-benchmark/ https://2026cnmllmtierlist.datacorp.top/en/articles/gui-agent-benchmark/ Wed, 14 Oct 2026 08:00:00 +0800 Everyone compares GUI-agent task success rates. On real devices over long tasks, what decides the outcome is whether the agent walks itself back after an error. We build a recovery-rate tier list over six failure modes and publish coordinate-error P50/P95 — together, the actual basis for the tier list’s agent dimension. en https://2026cnmllmtierlist.datacorp.top/en/articles/long-video-understanding/ https://2026cnmllmtierlist.datacorp.top/en/articles/long-video-understanding/ Tue, 13 Oct 2026 08:00:00 +0800 A single video score hides the thing that matters most: how accuracy decays as required time precision tightens from ±60s to ±1s. We introduce the temporal tolerance curve and its area, TTA, then re-rank all 32 models on video. The result differs from the tier list’s video dimension by up to six places — GLM-5.3 gains six, WITA-Omni four. en https://2026cnmllmtierlist.datacorp.top/en/articles/chinese-ocr-field-guide/ https://2026cnmllmtierlist.datacorp.top/en/articles/chinese-ocr-field-guide/ Sun, 11 Oct 2026 08:00:00 +0800 Public benchmarks only cover print, tables and handwriting. We built a separate mini tier list over eight document classes nobody tests — from vertical classics to tongue-diagnosis images, express waybills, mahjong tiles and industrial QC shots. These five are exactly where Chinese models hold a real moat and Western models do not compete at all. en https://2026cnmllmtierlist.datacorp.top/en/articles/benchmark-traps/ https://2026cnmllmtierlist.datacorp.top/en/articles/benchmark-traps/ Sat, 10 Oct 2026 08:00:00 +0800 Rather than listing seven ways leaderboards mislead, we quantify the gap. For all 32 models we compute a rank-divergence index: the distance between a model’s place on public boards like SuperCLUE-VLM and its place here. The five largest divergences have five different causes, and none of them is pay-to-win. en https://2026cnmllmtierlist.datacorp.top/en/articles/tierlist-methodology/ https://2026cnmllmtierlist.datacorp.top/en/articles/tierlist-methodology/ Thu, 08 Oct 2026 08:00:00 +0800 Every tier list is a function of its weights, so any ranking published without error bars is faking precision. We perturb the eight dimension weights by ±30%, rerun 2,000 times, compute each model’s rank variance, and split the 26 into “weight-robust” and “weight darlings” — with the full percentile table published. It is the only hard test of whether a ranking deserves trust. en https://2026cnmllmtierlist.datacorp.top/en/articles/native-multimodal-explained/ https://2026cnmllmtierlist.datacorp.top/en/articles/native-multimodal-explained/ Tue, 06 Oct 2026 08:00:00 +0800 We split “native multimodal” into five falsifiable grades, N0 to N4, place all 26 Chinese models, then rerun the ranking as a counterfactual: if this 2026 tier list admitted only N3 and above, who gets ejected and who jumps straight to the top. en