Chinese AI Model Rankings
Where the Chinese AI models ChinaAPI serves stand next to the international frontier, quoted from public leaderboards. Snapshot of 2026-08-10. Every figure is reproduced from the board named in its column and links back to it. ChinaAPI runs no evaluation of its own and computes no composite score from these numbers. See how this list is built, the price of every model marked available, and the same data as JSON at /api/model-rankings.
Sources
- LMArena Text Arena — LMArena (arena.ai), Arena score (Elo), live board, retrieved 2026-08-10
- LiveBench — LiveBench (sponsored by Abacus.AI), Overall, LiveBench-2026-06-25, retrieved 2026-08-10
- Artificial Analysis Intelligence Index — Artificial Analysis, Intelligence Index, live board, retrieved 2026-08-10
- SuperCLUE 智能指数 — SuperCLUE, 总分 (overall), 2026-07 evaluation, published 2026-08-06, retrieved 2026-08-10
- Epoch Capabilities Index (ECI) — Epoch AI, General ECI, live board, retrieved 2026-08-10, CC BY 4.0, as labelled on the page (https://creativecommons.org/licenses/by/4.0/)
- Vals Index — Vals AI, Accuracy (weighted finance and coding), V1.2, page states updated 8/9/2026, retrieved 2026-08-10
- Vendor's own evaluation card — the model's own vendor, varies; the benchmark is named in each row, self-reported, retrieved 2026-08-10
- LMArena Text-to-Image Arena — LMArena (arena.ai), Arena score (Elo), live board, retrieved 2026-08-10
- LMArena Image Edit Arena — LMArena (arena.ai), Arena score (Elo), live board, retrieved 2026-08-10
- LMArena Text-to-Video Arena — LMArena (arena.ai), Arena score (Elo), live board, retrieved 2026-08-10
- LMArena Image-to-Video Arena — LMArena (arena.ai), Arena score (Elo), live board, retrieved 2026-08-10
Text and multimodal models
Ordered by Epoch Capabilities Index (ECI); a model with no figure on that board follows the ones that have one, because absent from a board is not the same as last on it.
| # | Model | Vendor | Origin | On ChinaAPI | LMArena Text Arena | LiveBench | Artificial Analysis Intelligence Index | SuperCLUE 智能指数 | Epoch Capabilities Index (ECI) | Vals Index | Vendor's own evaluation card | Notes |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Claude Fable 5 | Anthropic | International | no | 1507 ±6 | 83.0 | 62 | — | 162 (159 - 165) | 75.14 | — | |
| 2 | GPT-5.6 Sol | OpenAI | International | no | 1482 ±6 | 81.0 | 61 | — | 162 (159 - 165) | 73.12 | — | |
| 3 | Claude Opus 5 | Anthropic | International | no | 1493 ±6 | 80.1 | 63 | — | 161 (159 - 164) | 74.82 | — | |
| 4 | GPT-5.5 | OpenAI | International | no | 1482 ±4 | 80.2 | — | — | 159 (157 - 162) | 67.95 | — | |
| 5 | GPT-5.6 Terra | OpenAI | International | no | 1467 ±6 | 77.9 | 57 | — | 159 (156 - 162) | 65.14 | — | |
| 6 | Kimi K3 | Moonshot AI | China | yes | 1485 ±7 | 79.2 | 60 | 70.68 | 157 (155 - 160) | 74.70 | — | |
| 7 | Qwen3.8-Max | Alibaba | China | yes | 1497 ±9 | 78.5 | 58 | 71.48 | 156 (154 - 158) | 65.47 | — | |
| 8 | Gemini 3.1 Pro | International | no | 1487 ±3 | 77.0 | 48 | — | 155 (153 - 157) | 53.77 | — | ||
| 9 | Gemini 3.6 Flash | International | yes | 1485 ±6 | 73.6 | 52 | — | 154 (152 - 156) | 62.43 | — | ||
| 10 | Grok 4.5 | xAI | International | no | 1468 ±6 | 75.8 | 56 | — | 154 (152 - 156) | 65.30 | — | LMArena labels this vendor SpaceXAI. |
| 11 | Qwen3.7-Max | Alibaba | China | yes | 1475 ±10 | 73.1 | — | — | 154 (152 - 156) | 57.49 | — | LMArena lists the preview build (qwen3.7-max-preview); the released model has no separate row. |
| 12 | DeepSeek V4 Flash | DeepSeek | China | yes | — | 74.2 | 52 | 65.60 | 153 (150 - 154) | 63.95 | — | The boards carry the 0731 build; LiveBench also lists an earlier DeepSeek V4 Flash row at 65.5. |
| 13 | GLM-5.2 | Zhipu AI (Z.ai) | China | yes | 1471 ±5 | 73.2 | 53 | 63.27 | 152 (151 - 154) | 65.02 | — | |
| 14 | GLM-5.1 | Zhipu AI (Z.ai) | China | yes | 1468 ±4 | — | — | — | 151 (149 - 152) | 52.45 | — | |
| 15 | Kimi K2.6 | Moonshot AI | China | yes | 1461 ±5 | 70.5 | — | — | 151 (149 - 153) | 55.17 | — | |
| 16 | Kimi K2.7 Code | Moonshot AI | China | yes | — | 68.4 | 43 | — | 151 (149 - 152) | — | — | |
| 17 | DeepSeek V4 Pro | DeepSeek | China | yes | 1457 ±4 | 71.6 | 45 | 64.40 | 149 (147 - 151) | 55.62 | — | The Vals Index labels this row "DeepSeek V4"; it lists the Flash build separately. |
| 18 | MiniMax-M3 | MiniMax | China | yes | — | 67.3 | 45 | 56.90 | 149 (145 - 151) | 58.94 | — | |
| 19 | Qwen3.6-Plus | Alibaba | China | yes | — | 68.9 | — | — | 149 (147 - 151) | 48.89 | — | |
| 20 | GLM-5 | Zhipu AI (Z.ai) | China | yes | 1457 ±4 | — | — | — | 146 (144 - 148) | — | — | |
| 21 | MiniMax-M2.7 | MiniMax | China | yes | — | — | — | — | 146 (138 - 148) | 41.56 | — | |
| 22 | Qwen3.6-Flash | Alibaba | China | yes | — | — | — | — | 144 (141 - 146) | — | — | |
| 23 | Agnes 2.0 Flash | Agnes AI | China | yes | — | — | — | — | — | — | Claw-Eval Pass^3 60.9% | No independent board in this snapshot covers it; the figure shown is the vendor's own. |
| 24 | Agnes 2.5 Flash | Agnes AI | China | yes | — | — | — | — | — | — | — | Artificial Analysis carries a different build of this family (Agnes 2.5 Pro Alpha, 40) and the vendor publishes its own figures only as a chart image, so there is no number here that can be quoted. |
| 25 | Doubao-Seed-2.1-pro | ByteDance | China | yes | — | — | — | 65.94 | — | — | Terminal Bench 2.1 71.0 | |
| 26 | Doubao-Seed-2.1-turbo | ByteDance | China | yes | — | — | — | — | — | — | Terminal Bench 2.1 67.6 | No independent board in this snapshot covers it; the figure shown is the vendor's own. The vendor writes the name without the 260628 snapshot suffix we serve it under. |
| 27 | GLM-5-Turbo | Zhipu AI (Z.ai) | China | yes | — | — | — | — | — | — | — | No independent board in this snapshot carries a row for it. |
| 28 | GLM-5V-Turbo | Zhipu AI (Z.ai) | China | yes | — | — | — | — | — | — | — | Vision-language model; no independent board in this snapshot carries a row for it. |
| 29 | Hunyuan Hy3 | Tencent | China | yes | 1453 ±10 | — | 42 | 62.13 | — | — | — | |
| 30 | LongCat-2.0 | Meituan | China | yes | — | — | 34 | 54.22 | — | — | — | |
| 31 | MiMo-V2.5 | Xiaomi | China | yes | — | — | 38 | — | — | 51.57 | — | |
| 32 | MiMo-V2.5-Pro | Xiaomi | China | yes | 1467 ±4 | — | 43 | 53.01 | — | 50.74 | — | |
| 33 | Muse Spark 1.2 | Meta | International | no | 1498 ±13 | 78.0 | 57 | — | — | 71.88 | — | |
| 34 | Qwen3.7-Plus | Alibaba | China | yes | 1458 ±5 | — | 39 | — | — | 52.33 | — | |
| 35 | Step 3.5 Flash | StepFun | China | yes | — | — | — | — | — | — | SWE-bench Verified 74.4 | No independent board in this snapshot covers it; the figure shown is the vendor's own. |
| 36 | Step 3.7 Flash | StepFun | China | yes | — | — | 31 | 48.16 | — | — | — |
Image generation models
Ordered by LMArena Text-to-Image Arena; a model with no figure on that board follows the ones that have one, because absent from a board is not the same as last on it.
| # | Model | Vendor | Origin | On ChinaAPI | LMArena Text-to-Image Arena | LMArena Image Edit Arena | Notes |
|---|---|---|---|---|---|---|---|
| 1 | GPT Image 2 | OpenAI | International | no | 1380 ±5 | 1463 ±4 | |
| 2 | Gemini 3.1 Flash Image | International | no | 1263 ±5 | 1385 ±4 | ||
| 3 | Doubao Seedream 5.0 | ByteDance | China | yes | 1257 ±5 | 1393 ±4 | Scores are the arena's seedream-5.0-pro rows; the arena also carries a seedream-5.0-lite build (1136 / 1293). |
| 4 | FLUX.2 Pro | Black Forest Labs | International | no | 1155 ±3 | 1244 ±3 | |
| 5 | Doubao Seedream 4.5 | ByteDance | China | yes | 1146 ±3 | 1301 ±2 | |
| 6 | Wan 2.7 Image Pro | Alibaba | China | yes | 1103 ±5 | 1302 ±4 | |
| 7 | Wan 2.7 Image | Alibaba | China | yes | 1100 ±5 | 1301 ±4 | |
| 8 | Agnes Image 2.0 Flash | Agnes AI | China | yes | — | — | No row on the LMArena image boards at the time of this snapshot. |
| 9 | Agnes Image 2.1 Flash | Agnes AI | China | yes | — | — | No row on the LMArena image boards at the time of this snapshot. |
| 10 | MiniMax image-01 | MiniMax | China | yes | — | — | No row on the LMArena image boards at the time of this snapshot. |
| 11 | Step 2X Large | StepFun | China | yes | — | — | No row on the LMArena image boards at the time of this snapshot; the arena carries an older StepFun build (step1x-edit, 998) that is a different model. |
| 12 | Step Image Edit 2 | StepFun | China | yes | — | — | No row on the LMArena image boards at the time of this snapshot; the arena carries an older StepFun build (step1x-edit, 998) that is a different model. |
Video generation models
Ordered by LMArena Text-to-Video Arena; a model with no figure on that board follows the ones that have one, because absent from a board is not the same as last on it.
| # | Model | Vendor | Origin | On ChinaAPI | LMArena Text-to-Video Arena | LMArena Image-to-Video Arena | Notes |
|---|---|---|---|---|---|---|---|
| 1 | Gemini Omni Flash (video) | International | no | 1513 ±12 | 1462 ±7 | ||
| 2 | Doubao Seedance 2.0 | ByteDance | China | yes | 1479 ±11 | 1478 ±10 | The arena entry is the 720p Dreamina build of Seedance 2.0. |
| 3 | MiniMax H3 | MiniMax | China | yes | 1455 ±19 | 1476 ±19 | |
| 4 | Sora 2 Pro | OpenAI | International | no | 1365 ±8 | — | |
| 5 | Veo 3.1 | International | no | 1364 ±14 | 1397 ±11 | ||
| 6 | Wan 2.7 T2V | Alibaba | China | yes | 1347 ±9 | — | |
| 7 | MiniMax Hailuo 2.3 | MiniMax | China | yes | 1202 ±7 | 1260 ±6 | |
| 8 | MiniMax Hailuo 02 | MiniMax | China | yes | 1198 ±13 | 1227 ±11 | Scores are the arena's hailuo-02-pro rows; the arena also carries standard and fast builds of the same family. |
| 9 | Doubao Seedance 2.5 | ByteDance | China | yes | — | — | No row on the LMArena video boards at the time of this snapshot. |
| 10 | HappyHorse 1.1 I2V | Alibaba | China | yes | — | — | The LMArena video boards carry the previous generation (happyhorse-1.0, 1428 text-to-video / 1442 image-to-video); that score is not this model's and is deliberately not reused here. |
| 11 | HappyHorse 1.1 T2V | Alibaba | China | yes | — | — | The LMArena video boards carry the previous generation (happyhorse-1.0, 1428 text-to-video / 1442 image-to-video); that score is not this model's and is deliberately not reused here. |
| 12 | Kling 3.0 Turbo | Kuaishou | China | yes | — | — | No row on the LMArena video boards at the time of this snapshot. |
| 13 | Kling v3 | Kuaishou | China | yes | — | 1359 ±7 | The arena entry is the pro build; the text-to-video board has no v3 row at the time of this snapshot. |
| 14 | Kling v3 Omni | Kuaishou | China | yes | — | — | No row on the LMArena video boards at the time of this snapshot. |
| 15 | MiniMax Hailuo 2.3 Fast | MiniMax | China | yes | — | — | No row on the LMArena video boards at the time of this snapshot. |
| 16 | Wan 2.7 I2V | Alibaba | China | yes | — | 1427 ±7 |
Real usage on ChinaAPI
Which models the calls ChinaAPI actually served went to, as positions and shares of this platform's own traffic. It is a measurement of one gateway, not of the Chinese AI market, and it is not comparable with the benchmark results above. No call count, token count or revenue figure is published here or anywhere else on this site: every number below is a proportion. Recomputed hourly. A model with too little traffic in a window to rank is left unnamed and counted under “other”, as is anything outside the published catalogue. The same board as JSON is at /api/model-rankings/usage.
Share of calls served, the last 7 days
| # | Model | Vendor | Share of calls | Trend vs the previous 7 days |
|---|---|---|---|---|
| 1 | deepseek-v4-flash | DeepSeek | 32.3% | flat |
| 2 | deepseek-v4-pro | DeepSeek | 25.1% | up |
| 3 | mimo-v2.5 | Xiaomi | 23.3% | up |
| 4 | glm-5.2 | 智谱 | 2.1% | down |
| 5 | hy3 | Tencent | 1.9% | up |
| 6 | kimi-k3 | Moonshot | 1.3% | down |
| 7 | qwen3.8-max | 阿里巴巴 | 1.3% | up |
| 8 | glm-5-turbo | 智谱 | 1.2% | up |
| 9 | glm-5 | 智谱 | 1.0% | flat |
| 10 | kimi-k2.7-code | Moonshot | 0.7% | flat |
| 11 | glm-5.1 | 智谱 | 0.6% | flat |
| 12 | kimi-k2.6 | Moonshot | 0.4% | flat |
| 13 | MiniMax-M3 | MiniMax | 0.3% | flat |
| All other models | 8.3% |
How those calls arrived: 0.2% over agent protocols (Claude Messages and the OpenAI Responses API), 8.1% over chat protocols (OpenAI chat completions and Gemini generateContent), 91.7% over everything else (audio, images, embeddings, rerank and asynchronous tasks).
Share of calls served, the last 30 days
| # | Model | Vendor | Share of calls | Trend vs the previous 30 days |
|---|---|---|---|---|
| 1 | deepseek-v4-flash | DeepSeek | 27.3% | up |
| 2 | deepseek-v4-pro | DeepSeek | 22.5% | up |
| 3 | mimo-v2.5 | Xiaomi | 18.4% | up |
| 4 | glm-5.2 | 智谱 | 7.5% | up |
| 5 | kimi-k3 | Moonshot | 5.9% | up |
| 6 | mimo-v2.5-pro | Xiaomi | 2.3% | up |
| 7 | hy3 | Tencent | 1.4% | down |
| 8 | MiniMax-M2.7 | MiniMax | 1.2% | up |
| 9 | qwen3.6-flash | 阿里巴巴 | 0.9% | up |
| 10 | glm-5 | 智谱 | 0.9% | up |
| 11 | glm-5.1 | 智谱 | 0.7% | up |
| 12 | glm-5-turbo | 智谱 | 0.5% | flat |
| 13 | qwen3.7-max | 阿里巴巴 | 0.5% | down |
| 14 | kimi-k2.7-code | Moonshot | 0.4% | flat |
| 15 | qwen3.8-max | 阿里巴巴 | 0.4% | flat |
| 16 | kimi-k2.6 | Moonshot | 0.4% | flat |
| 17 | MiniMax-M3 | MiniMax | 0.4% | flat |
| 18 | stepaudio-2.5-asr | StepFun | 0.2% | flat |
| 19 | qwen3.7-plus | 阿里巴巴 | 0.2% | flat |
| 20 | mimo-v2.5-asr | Xiaomi | 0.2% | flat |
| All other models | 7.8% |
How those calls arrived: 0.1% over agent protocols (Claude Messages and the OpenAI Responses API), 2.6% over chat protocols (OpenAI chat completions and Gemini generateContent), 97.3% over everything else (audio, images, embeddings, rerank and asynchronous tasks).