Chinese AI Model Rankings

As of 2026-09-22, Claude Opus 5.5 leads the Artificial Analysis Intelligence Index at 58; the highest-placed Chinese model is MiMo-V2.6-Pro at 46 (#10). 82 of the 100 models listed can be called through ChinaAPI.

Where the Chinese AI models ChinaAPI serves stand next to the international frontier, quoted from public leaderboards. Snapshot of 2026-09-22; roster last maintained 2026-09-22 (moves when a model or source is added). 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

Text and multimodal models

Ordered by Artificial Analysis Intelligence Index; 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.

Text and multimodal models, public benchmark results as of 2026-09-22
#ModelVendorOriginReleasedOn ChinaAPILMArena Text ArenaLiveBenchArtificial Analysis Intelligence IndexSuperCLUE 智能指数Epoch Capabilities Index (ECI)Vals IndexVendor's own evaluation cardLMArena Text Arena · ChineseLMArena Text Arena · CodingLiveBench · CodingLiveBench · Agentic CodingArtificial Analysis · Output SpeedArtificial Analysis · GDPval-AA v2.1Artificial Analysis · Terminal-Bench v4.0Artificial Analysis · MLCR-AANotes
1Claude Opus 5.5AnthropicInternational2026-09-22yes—83.258——66.16 ±1.00———89.371.7—6760—Released 2026-09-22 per Anthropic's announcement (Introducing Claude Opus 5.5), the first model in the Claude 5.5 family. Artificial Analysis printed no output speed for its Claude Opus 5.5 (max with fallback) row on 2026-09-22 (the column reads --), so no speed figure is quoted; the speeds of its lower-effort rows are not substituted.
2Claude Fable 5.1AnthropicInternational2026-09-01yes1498 ±883.453—165 (162 - 170)68.83 ±1.08—1592 ±321519 ±1886.466.165625271.1Released 2026-09-01 per Anthropic's announcement (Introducing Claude Fable 5.1 and Claude Mythos 5.1); on sale here since 2026-09-02. LMArena's row is the max-effort build, 5,783 votes as of 2026-09-22 and not marked preliminary; the same row on the Chinese category stands on 369 votes. Epoch's fit as read on 2026-09-22 scores it 165.00 and Claude Fable 5 163.60.
3GPT-6 AstraOpenAIInternational2026-09-03yes1480 ±1282.253—167 (163 - 172)66.61 ±1.09——1543 ±2380.457.358525935.0Released 2026-09-03 per OpenAI's announcement (GPT-6 Astra: A new generation of intelligence). LMArena added gpt-6-astra-max to its text board on 2026-09-11 per its leaderboard changelog; the row stands on 2,693 votes as of 2026-09-22, and the Chinese category carries no row for it. On sale here since 2026-09-05.
4Claude Opus 5AnthropicInternational2026-07-24yes1493 ±480.151—163 (160 - 167)67.21 ±0.98—1559 ±121533 ±781.465.254604955.6
5Claude Fable 5AnthropicInternational2026-06-09yes1506 ±583.050—164 (161 - 168)66.04 ±1.03—1553 ±141552 ±786.062.261554264.4Artificial Analysis marked its Claude Fable 5 (with fallback) row deprecated between the 2026-09-12 and 2026-09-21 reads; the leaderboard hides such rows under its default Status: Current filter, so this row's leaderboard figures are read with the Status filter set to All. LMArena relabelled this row from claude-fable-5 to claude-fable-5-high between 2026-09-06 and 2026-09-18: the Internet Archive's copies of the board show the same model key (claude-fable-5-text) with its vote count running on from 27,189 to 30,057, so the entry follows the new label on all three LMArena text boards.
6GPT-6 SolOpenAIInternational2026-09-22yes——48————————1044944—Released 2026-09-22 per OpenAI's announcement (Introducing GPT-6 Sol and Luna) and the API changelog entry of the same day.
7Muse Spark 1.3MetaInternational2026-09-02no1493 ±981.648—157 (155 - 159)60.31 ±1.08—1541 ±341537 ±1681.164.1219593343.3Released 2026-09-02 per Meta AI Research's announcement (Introducing Muse Spark 1.3). Epoch, which carried no row for it as of 2026-09-05, lists it as Muse Spark 1.3 (dated 2026-09-02) in the 2026-09-22 read. LMArena added muse-spark-1.3-max to its text board on 2026-09-13 per its leaderboard changelog; the row stands on 4,723 votes as of 2026-09-22. Not on sale here as of 2026-09-22.
8GPT-5.6 SolOpenAIInternational2026-07-09yes1483 ±581.047—162 (160 - 166)63.71 ±1.06—1539 ±151528 ±883.956.273544026.1OpenAI announced a limited partner preview of the GPT-5.6 series (Sol, Terra and Luna) on 2026-06-26 and made the models generally available on 2026-07-09, the post the preview page links as the launch; release_date follows the launch.
9Grok 4.7xAIInternational2026-09-21no——46——60.20 ±1.08—————39602615.0Released 2026-09-21 per the vendor's announcement, which now publishes under the SpaceXAI name; the row keeps the xAI label of the earlier Grok rows. Artificial Analysis lists Grok 4.7 (xhigh) and Grok 4.7 (high), both printing 46; the xhigh row is quoted because it is the highest effort setting and the higher score before rounding (46.4 against 46.3).
10MiMo-V2.6-ProXiaomiChina2026-09-22yes——46————————555935—Released and open-sourced 2026-09-22 per Xiaomi's release record (MiMo-V2.6 Series Release).
11GLM-5.3Zhipu AI (Z.ai)China2026-08-14yes1483 ±676.14571.29156 (154 - 158)56.97 ±1.35—1525 ±221524 ±1279.060.953574248.3Released 2026-08-14; served here via Zhipu's official channels since 2026-08-18. Independent boards picked it up between the 2026-08-19 and 2026-08-26 snapshots, so the vendor's own DeepSWE figure it carried before has been replaced by their readings; the LMArena row stands on 10,960 votes as of 2026-09-22. Epoch, which carried no row for it as of 2026-09-05, lists it as GLM-5.3 (dated 2026-08-14) in the 2026-09-22 read, and SuperCLUE lists GLM-5.3(max) with an evaluation published 2026-09-10. Vals lists it as GLM 5.3, released 2026-08-18 per its model page, the date of Z.ai's release notes.
12Qwen3.8-Max-0902AlibabaChina2026-09-02yes——4572.62155 (153 - 157)——————39583920.0Alibaba's 2026-09-02 snapshot of Qwen3.8-Max (alias qwen3.8-max-2026-09-02), listed on Model Studio on 2026-09-02 as an upgraded snapshot of qwen3.8-max; Epoch dates it 2026-09-01. Alibaba Cloud's update notices of 2026-09-02 (www.aliyun.com/notice/118616 and the international Model Studio notice Update Notice for Qwen3.8-Max Models) state that from 10:00 Beijing time on 2026-09-05, subject to the actual change time, the qwen3.8-max model name automatically moved to this snapshot with billing unchanged. Alibaba published no completion notice; Artificial Analysis has since marked the earlier build deprecated. This row therefore links to qwen3.8-max, the name on sale here; the same snapshot is also sold under its pinned name qwen3.8-max-0902. Artificial Analysis and Epoch list it as Qwen3.8 Max (0902), SuperCLUE as Qwen3.8-Max-0902(max) (evaluation published 2026-09-10). LMArena added qwen3.8-max-0902 only to its Code Arena (changelog, 2026-09-01), which this page does not cite, and neither LiveBench nor Vals carries a 0902 row; the 2026-08-02 build's figures on the qwen3.8-max row are not reused here.
13Grok 4.6xAIInternational2026-08-12no1456 ±678.044—156 (155 - 159)59.17 ±1.21—1517 ±181506 ±1076.857.058552112.2Released 2026-08-12. LMArena labels this vendor SpaceXAI. Its row carried the Preliminary badge on 3,453 votes as of 2026-09-05 and no longer does (15,521 votes as of 2026-09-22).
14Kimi K3Moonshot AIChina2026-07-16yes1485 ±579.24470.68158 (155 - 161)57.81 ±1.06—1535 ±161538 ±981.462.237511338.3
15Step 5 PreviewStepFunChina2026-09-20yes——44————————83533316.7StepFun's announcement page prints no date; it says the model is released today and cites Artificial Analysis results published as of 2026-09-20, and reports dated that day already quote the release, which fixes the release date. The vendor states open weights follow on 2026-10-15.
16GLM-5.3-FlashZhipu AI (Z.ai)China2026-08-26yes1475 ±771.64268.10152 (150 - 154)47.22 ±1.45—1529 ±251525 ±1279.056.865573351.1Released 2026-08-26 per Z.ai's own release notes; the first natively multimodal model in the GLM-5 line and its low-cost tier, served here on Zhipu's official channels. LMArena, LiveBench and Artificial Analysis all picked it up within days of release; the LMArena row is a ranked one standing on 10,038 votes as of 2026-09-22, carrying neither the board's Preliminary badge nor an AutoEval marker. Epoch, SuperCLUE and Vals carried no row for it as of 2026-08-31. In the 2026-09-22 read Epoch lists it as GLM-5.3-Flash dated 2026-08-20, six days before Z.ai's release notes; the benchmark runs behind that figure are on the glm-5.3-flash model itself (Epoch's model versions glm-5.3-flash_max and glm-5.3-flash_high), so only the date differs. SuperCLUE lists GLM-5.3-Flash(max) with an evaluation published 2026-09-10, and Vals lists it as GLM 5.3 Flash, released 2026-08-26 per its model page. GLM-5.3's figures are not reused here.
17GPT-5.6 TerraOpenAIInternational2026-07-09yes1466 ±577.942—159 (157 - 162)59.59 ±1.33—1516 ±141518 ±878.254.982473531.7OpenAI announced a limited partner preview of the GPT-5.6 series (Sol, Terra and Luna) on 2026-06-26 and made the models generally available on 2026-07-09, the post the preview page links as the launch; release_date follows the launch.
18Gemini 3.8 FlashGoogleInternational2026-09-02yes1493 ±975.841—157 (155 - 161)62.25 ±1.01—1543 ±321535 ±1672.554.2297462021.7Released 2026-09-02 per Google's announcement (Introducing Gemini 3.8 Flash and 3.8 Flash Cyber). LMArena marks the row preliminary (5,076 votes as of 2026-09-22). Epoch, which carried no row for it as of 2026-09-05, lists it as Gemini 3.8 Flash (dated 2026-09-02) in the 2026-09-22 read. Confirmed on sale in the public catalogue read on 2026-09-12.
19Muse Spark 1.2MetaInternational2026-08-05no1500 ±1178.040—155 (153 - 158)57.05 ±1.12—1528 ±381536 ±2077.557.617249731.1Epoch added a Muse Spark 1.2 row (dated 2026-08-05) between the 2026-08-31 and 2026-09-05 snapshots. Artificial Analysis marked its Muse Spark 1.2 (xhigh) row deprecated between the 2026-09-12 and 2026-09-21 reads; the leaderboard hides such rows under its default Status: Current filter, so this row's leaderboard figures are read with the Status filter set to All.
20Qwen3.8-2.4T-A95BAlibabaChina—yes——40————————3855110.0The open-weight Qwen3.8-2.4T-A95B release has its own Artificial Analysis row. Qwen identifies Qwen3.8-Max as a production version based on this model with additional features; the two catalogue models remain separate here, and Max scores from other boards are not reused. Artificial Analysis prints MLCR-AA 0.0 for this release, with no breakdown, against 19.4 and 20.0 for the two Qwen3.8-Max builds.
21Qwen3.8-FlashAlibabaChina2026-08-26yes—76.24066.41—————72.661.6545625—The boards list this model under its open-weight release name, Qwen3.8-Flash-Next (released 2026-08-26 per the Qwen team). The vendor's announcement states that the production version, with a 1M default context and built-in tools, is served under the name Qwen3.8-Flash, which is the qwen3.8-flash API on sale here since 2026-09-04; the two rows are joined on that statement, not on the row name. Artificial Analysis now prints 40 without an estimated marker (read 2026-09-21). SuperCLUE lists it under the API name itself, Qwen3.8-Flash(max), evaluated through the API with an evaluation published 2026-09-10, so that figure needs no join. Neither Epoch, LMArena nor Vals carried a row as of 2026-09-05.
22Qwen3.8-MaxAlibabaChina2026-08-02no1481 ±678.54071.48157 (155 - 159)51.84 ±1.29—1538 ±181522 ±972.964.638551919.4The 2026-08-02 build of Qwen3.8-Max (Artificial Analysis dates it 2026-08-03), which replaced Qwen3.8-Max-Preview (retired 2026-08-05). Its release_date follows Epoch, which re-dated this build from 2026-07-19, the Preview's announcement at WAIC, to 2026-08-02 between its 2026-08-25 and 2026-09-03 releases; Model Studio listed qwen3.8-max on 2026-08-02 as well. Alibaba Cloud's update notices of 2026-09-02 (www.aliyun.com/notice/118616 and the international Model Studio notice Update Notice for Qwen3.8-Max Models) state that from 10:00 Beijing time on 2026-09-05, subject to the actual change time, the qwen3.8-max model name automatically moved to the qwen3.8-max-0902 snapshot with billing unchanged. The name on sale here therefore now serves the Qwen3.8-Max-0902 row, and Model Studio lists no snapshot ID for this build, so it can no longer be called there. Every figure on this row belongs to this build: Artificial Analysis's Qwen3.8 Max (slug qwen3-8-max-0803), Epoch's Qwen 3.8 Max (dated 2026-08-02), SuperCLUE's Qwen3.8-Max(max) (evaluation published 2026-08-06), LMArena's qwen3.8-max (added to its Text Arena on 2026-08-02 per its changelog), LiveBench's Qwen 3.8 Max (quoted at the same figures since 2026-08-10, before the snapshot existed) and Vals's Qwen 3.8 Max (released 2026-08-03 per its model page). Artificial Analysis marked its Qwen3.8 Max row deprecated between the 2026-09-12 and 2026-09-21 reads; the leaderboard hides such rows under its default Status: Current filter, so this row's leaderboard figures are read with the Status filter set to All.
23DeepSeek V4.1 FlashDeepSeekChina2026-09-10yes—81.13971.81155 (149 - 158)57.86 ±1.16———80.077.3220552722.8Released 2026-09-10 per DeepSeek's official announcement and served by the canonical deepseek-flash API model. The vendor retired V4 Flash and V4 Flash Vision Exp that day and temporarily routes their old identifiers to V4.1 Flash; their historical leaderboard rows remain separate and none of their figures are reused here. Artificial Analysis, LiveBench and Vals all published distinct V4.1 Flash rows by 2026-09-12; Epoch (DeepSeek V4.1 Flash, dated 2026-09-09) and SuperCLUE (DeepSeek-V4.1-Flash(max), evaluation published 2026-09-11) carry one as of the 2026-09-22 read.
24Gemini 3.7 FlashGoogleInternational2026-08-13yes1490 ±878.839—158 (156 - 161)59.31 ±1.06—1556 ±301521 ±1578.958.3281441415.0Released 2026-08-13. LMArena marks the row preliminary (5,640 votes as of 2026-09-22). Artificial Analysis marked its Gemini 3.7 Flash (high) row deprecated between the 2026-09-12 and 2026-09-21 reads; the leaderboard hides such rows under its default Status: Current filter, so this row's leaderboard figures are read with the Status filter set to All.
25Grok 4.5xAIInternational2026-07-08no1468 ±575.839—154 (152 - 156)51.53 ±1.32—1512 ±141518 ±768.656.555441115.0LMArena labels this vendor SpaceXAI. Artificial Analysis marked its Grok 4.5 (high) row deprecated between the 2026-09-12 and 2026-09-21 reads; the leaderboard hides such rows under its default Status: Current filter, so this row's leaderboard figures are read with the Status filter set to All. Released 2026-07-08 per xAI's announcement as published that day and its API release notes; the live announcement page now shows Jul 16, the date it was edited when the model reached the EU.
26GPT-5.6 LunaOpenAIInternational2026-07-09yes1452 ±573.637—156 (154 - 159)59.88 ±1.09—1476 ±141498 ±882.948.4144471219.4The low-cost tier of the GPT-5.6 line. OpenAI announced a limited partner preview of the GPT-5.6 series (Sol, Terra and Luna) on 2026-06-26 and made the models generally available on 2026-07-09, the post the preview page links as the launch; release_date follows the launch.
27GPT-6 LunaOpenAIInternational2026-09-22yes——37————————1574313—Released 2026-09-22 per OpenAI's announcement (Introducing GPT-6 Sol and Luna) and the API changelog entry of the same day.
28Agnes 3.0 FlashAgnes AIChina—yes——36*—————————537—Artificial Analysis lists Agnes 3.0 Flash separately from the 2.5 family. Its Intelligence Index is currently estimated, so the trailing * is retained exactly as the board prints it. Artificial Analysis no longer printed an output-speed value for this row on 2026-09-21 (the column reads --), so the 238 tokens/s figure was removed rather than retained without a live source; the index, GDPval-AA and Terminal-Bench figures are from the 2026-09-21 board read.
29DeepSeek V4 Pro 0813DeepSeekChina2026-08-13yes1463 ±777.43670.10155 (154 - 157)52.37 ±1.14—1494 ±251503 ±1277.254.968471417.8DeepSeek's change log dates this build's GA release to 2026-08-13 and states that the deepseek-v4-pro model name now serves it. Epoch added a DeepSeek V4 Pro 0813 row between the 2026-08-26 and 2026-08-31 snapshots, so the undated DeepSeek-V4-Pro figure is no longer borrowed for it. LMArena's 0813 row stands on 9,008 human votes as of 2026-09-22. SuperCLUE announced on 2026-08-26 that it had added the official DeepSeek-V4-Pro; in the July 2026 edition's data file read on 2026-09-22 that row is labelled DeepSeek-V4-Pro-0813(max), with an evaluation published 2026-08-18, and the 64.40 row the board had labelled DeepSeek-V4-Pro(max) 预览版, now plain DeepSeek-V4-Pro(max), is the preview build, quoted on the deepseek-v4-pro row rather than here.
30DeepSeek V4 Flash Vision ExpDeepSeekChina2026-08-21no—76.835——————68.265.12315212—DeepSeek's experimental vision tier, announced in the vendor's change log on 2026-08-21 as model='deepseek-v4-flash-vision-exp'. LiveBench names its row with the vendor's -Exp suffix and Artificial Analysis drops it, but DeepSeek publishes only one vision tier, so both rows are this model. LMArena, Vals, SuperCLUE and Epoch carry no row for it, and the sibling flash figures are not reused. DeepSeek retired this tier on 2026-09-10 and its compatibility ID now routes to V4.1 Flash. It is absent from the public catalogue read on 2026-09-12 and is retained here as a historical evaluation row without an on-sale badge.
31DeepSeek V4 Flash 0731DeepSeekChina2026-07-31no—74.23465.60155 (152 - 156)53.57 ±1.21———75.046.8221461213.3Historical 2026-07-31 V4 Flash build. DeepSeek retired V4 Flash on 2026-09-10 and its compatibility ID now routes to V4.1 Flash; this historical score row is not the currently served build and is not marked on sale. Artificial Analysis marked its DeepSeek V4 Flash 0731 (max) row deprecated between the 2026-09-12 and 2026-09-21 reads; the leaderboard hides such rows under its default Status: Current filter, so this row's leaderboard figures are read with the Status filter set to All.
32GLM-5.2Zhipu AI (Z.ai)China2026-06-16yes1472 ±573.23463.27152 (150 - 154)53.12 ±1.18—1517 ±131510 ±779.751.8684317.2Artificial Analysis marked its GLM-5.2 (max) row deprecated between the 2026-09-12 and 2026-09-21 reads; the leaderboard hides such rows under its default Status: Current filter, so this row's leaderboard figures are read with the Status filter set to All.
33Gemini 3.6 FlashGoogleInternational2026-07-21yes1480 ±573.634—154 (153 - 156)55.35 ±1.09—1536 ±151517 ±877.943.418038714.4Artificial Analysis marked its Gemini 3.6 Flash row deprecated between the 2026-09-12 and 2026-09-21 reads; the leaderboard hides such rows under its default Status: Current filter, so this row's leaderboard figures are read with the Status filter set to All.
34Qwen3.8-27BAlibabaChina2026-08-14yes1437 ±675.334—149 (147 - 151)48.48 ±1.45—1496 ±231499 ±1175.761.44145621.7Open-weight Qwen3.8 model, released 2026-08-14 per the Qwen team's release notes (Hugging Face and ModelScope); on sale here since 2026-09-04. Epoch, which carried no row for it as of 2026-09-05, lists it as Qwen 3.8 27B (dated 2026-08-14) in the 2026-09-22 read. Artificial Analysis lists several effort settings; the first-listed xhigh row is quoted rather than a lower-effort sibling.
35DeepSeek V4 ProDeepSeekChina2026-04-24no1457 ±4—3064.40149 (147 - 151)42.89 ±1.19—1493 ±121501 ±6——86321521.1The pre-0813 build; the deepseek-v4-pro API moved to DeepSeek-V4-Pro-0813 on 2026-08-12. LiveBench dropped this row in favour of the dated builds between the 2026-08-19 and 2026-08-26 snapshots, so its 71.6 is no longer quoted. SuperCLUE's DeepSeek-V4-Pro(max) row (64.40), which its July 2026 edition labelled 预览版, is quoted here from CA-971 on, overturning CA-605. CA-605 left it out because LMArena lists deepseek-v4-pro and deepseek-v4-pro-high-preview separately, but LMArena's own data gives those two rows the same checkpoint key (deepseek-v4-pro-ch1-text and deepseek-v4-pro-ch1-thinking-text) and links both to DeepSeek's 2026-04-24 announcement: they are one build in two modes, which DeepSeek's change log calls V4-Pro-Preview. Before the 2026-08-13 GA that build was the only one the deepseek-v4-pro API served, so an API evaluation published in the July edition can only be this row. SuperCLUE's data file has since dropped the 预览版 label and dates the row 2026.9.10, keeping the July figure. Artificial Analysis marked its DeepSeek V4 Pro (max) row deprecated between the 2026-09-12 and 2026-09-21 reads; the leaderboard hides such rows under its default Status: Current filter, so this row's leaderboard figures are read with the Status filter set to All.
36Gemini 3.1 ProGoogleInternational2026-02-19no1487 ±377.030—155 (153 - 158)41.90 ±1.17—1530 ±81520 ±576.544.111514415.6
37MiniMax-M3MiniMaxChina2026-06-01yes1441 ±467.32956.90147 (143 - 150)42.72 ±1.20—1471 ±121496 ±668.240.710637217.2
38Kimi K2.7 CodeMoonshot AIChina2026-06-12yes—68.426—150 (148 - 152)————74.045.76126116.1
39MiMo-V2.5-ProXiaomiChina2026-04-22yes1467 ±4—2653.01—40.97 ±1.38—1513 ±111521 ±6——453009.4
40Hunyuan Hy3TencentChina2026-07-06yes1456 ±7—2562.13———1520 ±251503 ±13——87271—
41MiMo-V2.5XiaomiChina2026-04-22yes1434 ±4—25*——39.91 ±1.31—1483 ±131491 ±7——34240—
42Qwen3.7-PlusAlibabaChina2026-06-01yes1456 ±4—25—147 (146 - 149)38.65 ±1.17—1506 ±131501 ±7——651319.4Released 2026-06-01 (10:00 China time) per the Qwen blog, the day Model Studio also lists it; Epoch records 2026-06-02, the China-time day of the launch post on X.
43LongCat-2.0MeituanChina2026-06-30yes——1954.22————————180—Artificial Analysis no longer printed an output-speed value for this row on 2026-09-12, so the old figure was removed rather than retained without a live source.
44Step 3.7 FlashStepFunChina2026-05-29yes——19*48.16———————19317——
45Step 3.5 FlashStepFunChina2026-02-02yes1394 ±4—17*————1435 ±111450 ±6——126———StepFun's platform documentation lists step-3.5-flash as the base version and step-3.5-flash-2603 as an agent-optimised variant built from it; the 2603 build shipped on 2026-04-02 and has its own row here. LMArena added step-3.5-flash to its text board on 2026-02-10, linking the original open weights, seven weeks before 2603 shipped, so its three text-board rows are this build (57,137 votes on the overall board as of 2026-09-22). Artificial Analysis lists this build as Step 3.5 Flash (released 2026-02-02 on its model page) and marks the row deprecated; the leaderboard hides such rows under its default Status: Current filter, so its figures are read with the Status filter set to All, and its Intelligence Index is an estimate (*). The vendor's own SWE-bench Verified figure (74.4) that stood here while no independent board was quoted has been removed. LiveBench, Epoch, Vals and SuperCLUE carry no row for it.
46Step 3.5 Flash 2603StepFunChina2026-04-02yes——17*————————118———StepFun's agent-optimised variant of step-3.5-flash, whose reasoning_effort takes only low and high. Released 2026-04-02 per StepFun's announcement on its X account ("Step 3.5 Flash 2603 just shipped!"). Artificial Analysis lists it as Step 3.5 Flash 2603 (released 2026-04-02 on its model page) and marks the row deprecated; the leaderboard hides such rows under its default Status: Current filter, so its figures are read with the Status filter set to All, and its Intelligence Index is an estimate (*). LMArena carries no 2603 row: its step-3.5-flash row is the base build and is not reused here. LiveBench, Epoch, Vals and SuperCLUE carry no row for it either.
47GPT-5.5OpenAIInternational2026-04-23yes1482 ±480.2——159 (157 - 162)57.41 ±1.15—1515 ±101520 ±682.154.0————
48Kimi K2.6Moonshot AIChina2026-04-20yes1460 ±570.5——151 (149 - 153)43.47 ±1.17—1526 ±141514 ±778.646.9————
49Qwen3.6-PlusAlibabaChina2026-04-01yes1443 ±468.9——148 (145 - 150)31.98 ±0.98—1477 ±131495 ±678.241.4————
50Qwen3.7-MaxAlibabaChina2026-05-20yes1473 ±1073.1——154 (152 - 156)44.77 ±1.08—1518 ±381525 ±1874.243.6————LMArena lists the preview build (qwen3.7-max-preview); the released model has no separate row. Released 2026-05-20 (10:00 China time) per the Qwen blog; Epoch records 2026-05-19, and the blog itself renders the date in the reader's time zone, so US readers see the 19th.
51MiniMax-M2.7MiniMaxChina2026-03-18yes1415 ±4———146 (138 - 148)24.56 ±0.93—1443 ±101481 ±5——————
52GLM-5Zhipu AI (Z.ai)China2026-02-12yes1458 ±4———146 (144 - 148)——1515 ±151498 ±7——————Released 2026-02-12 per Z.ai's blog (GLM-5: From Vibe Coding to Agentic Engineering) and both of its release notes; Epoch records 2026-02-11, the US-time day of the launch post.
53GLM-5.1Zhipu AI (Z.ai)China2026-04-07yes1466 ±4———150 (148 - 152)——1516 ±121513 ±6——————The Vals Index V2 (2026-08-13) no longer lists this model; the V1.2 figure it once carried was removed with that revision.
54DeepSeek V4 FlashDeepSeekChina—no1436 ±4——————1472 ±121483 ±6——————The pre-0731 build. LMArena still lists it alongside 0731; LiveBench dropped its row between the 2026-08-19 and 2026-08-26 snapshots, so its 65.5 is no longer quoted.
55GLM-5V-TurboZhipu AI (Z.ai)China—yes1433 ±7——————1479 ±241489 ±12——————Vision-language model. LMArena's text board now carries a glm-5v-turbo row (9,361 votes as of 2026-09-22); Epoch does not score it.
56Doubao-Seed-2.1-proByteDanceChina2026-06-28yes———65.94——Terminal Bench 2.1 71.0————————
57Hy4-previewTencentChina2026-08-28yes—————55.40 ±1.25—————————Tencent announced and open-sourced Hy4 preview on 2026-08-28. The Vals Index lists it as Hy4 Preview (tencent/hy4-preview); Artificial Analysis, LMArena, LiveBench and Epoch carried no row for it as of 2026-09-21, and those figures are left blank. Hy3 scores and AutoEval results are not attributed to this row.
58Qwen3.6-FlashAlibabaChina2026-04-26yes————143 (141 - 145)——————————
59Agnes 2.0 FlashAgnes AIChina—no——————Claw-Eval Pass^3 60.9%————————No independent board in this snapshot covers it; the figure shown is the vendor's own. Retired from sale 2026-09-03 (CA-654): the vendor deprecated it in favour of agnes-2.5-flash; kept as a ranked row with on_chinaapi false so the roster count and maintained.at still describe the 2026-08-31 board read.
60Agnes 2.5 FlashAgnes AIChina—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.
61Doubao-Seed-2.1-turboByteDanceChina2026-06-28yes——————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.
62Doubao-Seed-EvolvingByteDanceChina—yes———————————————Doubao publishes Seed Evolving as a rolling model under the stable doubao-seed-evolving API ID, so capabilities can change without a dated version suffix. No independent score could be verified for this exact model on the boards cited by this snapshot as of 2026-09-12; Seed 2.1 scores are not reused.
63GLM-5-TurboZhipu AI (Z.ai)China2026-03-16yes———————————————No independent board in this snapshot carries a row for it.
64MiMo-V2.6-FlashXiaomiChina2026-09-22yes——————DeepSWE v1.1 65.7————————Released and open-sourced 2026-09-22 per Xiaomi's release record (MiMo-V2.6 Series Release). No independent board in this snapshot covers it; the figure shown is the vendor's own.
65MiMo-V2.6-Pro-UltraspeedXiaomiChina2026-09-22yes———————————————Xiaomi's high-speed serving mode of MiMo-V2.6-Pro, released 2026-09-22. The vendor states flagship V2.6-Pro performance at up to 20x the output speed but publishes no separate figures, and no board lists it separately, so MiMo-V2.6-Pro's scores are not reused here.

Image generation models

Quality coverage: 13/17 models, 4 boards from 2 evaluation operators. Public preference evidence, not comprehensive capability certification or a ChinaAPI endpoint test. Tasks and audio pools are separate; scores across columns are not interchangeable. Overlapping intervals do not establish a significant lead.

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.

Image generation models, public benchmark results as of 2026-09-22
#ModelVendorOriginReleasedOn ChinaAPILMArena Text-to-Image ArenaLMArena Image Edit ArenaArtificial Analysis Text To ImageArtificial Analysis EditingNotes
1GPT Image 2.5 SunburstOpenAIInternational2026-09-08yes1421 ±13 — gpt-image-2.5-sunburst; text_to_image; 95% CI; samples 3149; Preliminary; retrieved 2026-09-121520 ±9 — gpt-image-2.5-sunburst; image_editing; 95% CI; samples 6704; Preliminary; retrieved 2026-09-121179 -11/11 — GPT Image 2.5 Sunburst (max); text_to_image; 95% CI; samples 5159; retrieved 2026-09-131167 -10/10 — GPT Image 2.5 Sunburst (max); image_editing; 95% CI; samples 5216; retrieved 2026-09-13Both Arena rows are marked Preliminary as read on 2026-09-12: 3,149 text-to-image and 6,704 image-edit votes. Sunburst and Flare are distinct builds; their scores are not interchangeable.
2GPT Image 2.5 FlareOpenAIInternational2026-09-08yes1399 ±13 — gpt-image-2.5-flare; text_to_image; 95% CI; samples 2856; Preliminary; retrieved 2026-09-121491 ±9 — gpt-image-2.5-flare; image_editing; 95% CI; samples 5676; Preliminary; retrieved 2026-09-121187 -11/11 — GPT Image 2.5 Flare (max); text_to_image; 95% CI; samples 5236; retrieved 2026-09-131145 -10/10 — GPT Image 2.5 Flare (max); image_editing; 95% CI; samples 5319; retrieved 2026-09-13Both Arena rows are marked Preliminary as read on 2026-09-12: 2,856 text-to-image and 5,676 image-edit votes. Flare and Sunburst are distinct builds; their scores are not interchangeable.
3GPT Image 2OpenAIInternational2026-04-21yes1381 ±4 — gpt-image-2 (medium); text_to_image; 95% CI; samples not captured; retrieved 2026-09-121461 ±3 — gpt-image-2 (medium); image_editing; 95% CI; samples not captured; retrieved 2026-09-121171 -9/9 — GPT Image 2 (high); text_to_image; 95% CI; samples 15419; retrieved 2026-09-131114 -9/9 — GPT Image 2 (high); image_editing; 95% CI; samples 13425; retrieved 2026-09-13Arena quotes medium quality; Artificial Analysis quotes high quality. These are different test configurations, not interchangeable scores.
4Gemini 3.1 Flash ImageGoogleInternational—yes1261 ±5 — gemini-3.1-flash-image (nano-banana-2) [web-search]; text_to_image; 95% CI; samples not captured; retrieved 2026-09-121387 ±4 — gemini-3.1-flash-image (nano-banana-2) [web-search]; image_editing; 95% CI; samples not captured; retrieved 2026-09-121122 -8/8 — Nano Banana 2 (Gemini 3.1 Flash Image); text_to_image; 95% CI; samples 16711; retrieved 2026-09-131103 -9/9 — Nano Banana 2 (Gemini 3.1 Flash Image); image_editing; 95% CI; samples 13522; retrieved 2026-09-13Arena identifies a web-search-enabled entry; Artificial Analysis does not identify that condition in its row label. Do not assume identical configurations.
5Doubao Seedream 5.0 ProByteDanceChina—yes1257 ±4 — seedream-5.0-pro; text_to_image; 95% CI; samples not captured; retrieved 2026-09-121394 ±4 — seedream-5.0-pro; image_editing; 95% CI; samples not captured; retrieved 2026-09-121081 -8/8 — Seedream 5.0 Pro; text_to_image; 95% CI; samples 12814; retrieved 2026-09-131100 -9/9 — Seedream 5.0 Pro; image_editing; 95% CI; samples 12053; retrieved 2026-09-13The 260628 Pro SKU is distinct from the 260128 Lite SKU. Scores quote the arena's seedream-5.0-pro rows, not seedream-5.0-lite.
6Qwen Image 3.0 ProAlibabaChina—yes1254 ±7 — qwen-image-3.0-pro; text_to_image; 95% CI; samples not captured; retrieved 2026-09-12—1084 -10/10 — Qwen-Image-3.0-Pro; text_to_image; 95% CI; samples 5896; retrieved 2026-09-131076 -9/9 — Qwen-Image-3.0-Pro; image_editing; 95% CI; samples 12051; retrieved 2026-09-13No Qwen Image 3.0 Pro row on the Arena image-edit board as read on 2026-09-12. The qwen-image-2.0-pro-2026-06-22 row is an older build and is not reused here.
7FLUX.2 ProBlack Forest LabsInternational2025-11-25no1154 ±3 — flux-2-pro; text_to_image; 95% CI; samples not captured; retrieved 2026-09-121245 ±3 — flux-2-pro; image_editing; 95% CI; samples not captured; retrieved 2026-09-121009 -8/8 — FLUX.2 [pro]; text_to_image; 95% CI; samples 9794; retrieved 2026-09-131006 -7/7 — FLUX.2 [pro]; image_editing; 95% CI; samples 8015; retrieved 2026-09-13
8Doubao Seedream 4.5ByteDanceChina2025-11-28yes1147 ±3 — seedream-4.5; text_to_image; 95% CI; samples not captured; retrieved 2026-09-121302 ±2 — seedream-4.5; image_editing; 95% CI; samples not captured; retrieved 2026-09-121008 -9/9 — Seedream 4.5; text_to_image; 95% CI; samples 6308; retrieved 2026-09-131035 -10/10 — Seedream 4.5; image_editing; 95% CI; samples 4166; retrieved 2026-09-13
9Doubao Seedream 5.0 LiteByteDanceChina2026-01-28yes1138 ±3 — seedream-5.0-lite; text_to_image; 95% CI; samples not captured; retrieved 2026-09-121294 ±3 — seedream-5.0-lite; image_editing; 95% CI; samples not captured; retrieved 2026-09-12999 -9/9 — Seedream 5.0 Lite; text_to_image; 95% CI; samples 5027; retrieved 2026-09-131050 -10/10 — Seedream 5.0 Lite; image_editing; 95% CI; samples 4912; retrieved 2026-09-13The 260128 API SKU is Seedream 5.0 Lite. Scores quote the arena's seedream-5.0-lite rows; the distinct 260628 Pro SKU has its own row. The original rank anchor is preserved.
10Wan 2.7 Image ProAlibabaChina2026-04-01yes1103 ±5 — wan2.7-image-pro; text_to_image; 95% CI; samples not captured; retrieved 2026-09-121302 ±4 — wan2.7-image-pro; image_editing; 95% CI; samples not captured; retrieved 2026-09-12980 -9/9 — Wan 2.7 Pro; text_to_image; 95% CI; samples 5059; retrieved 2026-09-131039 -10/10 — Wan 2.7 Pro; image_editing; 95% CI; samples 4833; retrieved 2026-09-13
11Wan 2.7 ImageAlibabaChina2026-04-01yes1100 ±5 — wan2.7-image; text_to_image; 95% CI; samples not captured; retrieved 2026-09-121301 ±4 — wan2.7-image; image_editing; 95% CI; samples not captured; retrieved 2026-09-12977 -9/9 — Wan 2.7; text_to_image; 95% CI; samples 5029; retrieved 2026-09-131033 -9/9 — Wan 2.7; image_editing; 95% CI; samples 6291; retrieved 2026-09-13
12Step Image Edit 2StepFunChina—no——810 -10/10 — Step Image Edit 2; text_to_image; 95% CI; samples 2593; retrieved 2026-09-131008 -10/10 — Step Image Edit 2; image_editing; 95% CI; samples 4235; retrieved 2026-09-13No 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. Artificial Analysis evidence is now quoted separately for the identified task and exact entry.
13Agnes Image 2.0 FlashAgnes AIChina—yes———1017 -10/10 — Agnes Image 2.0 Flash; image_editing; 95% CI; samples 3713; retrieved 2026-09-13No row on the LMArena image boards at the time of this snapshot. Artificial Analysis evidence is now quoted separately for the identified task and exact entry.
14Agnes Image 2.1 FlashAgnes AIChina—yes————No row on the LMArena image boards at the time of this snapshot.
15Agnes Image 2.5 FlashAgnes AIChina—yes————No Agnes Image 2.5 Flash row on either Arena image board as read on 2026-09-12; no independent score is substituted.
16MiniMax image-01MiniMaxChina—yes————No row on the LMArena image boards at the time of this snapshot.
17Step 2X LargeStepFunChina—no————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.

Third-party price and generation-time references

Not ChinaAPI prices or endpoint measurements. These references never affect quality coverage or ranking. Compare only matching tasks and configurations. Generation time is a trailing-three-day median of successful requests including download, not retry overhead.

ModelTaskMetricValueEvidence
GPT Image 2.5 Sunbursttext_to_imageprice210.7 USD / 1,000 imagesGPT Image 2.5 Sunburst (max) — OpenAI; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
GPT Image 2.5 Sunburstimage_editingprice210.7 USD / 1,000 imagesGPT Image 2.5 Sunburst (max) — OpenAI; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
GPT Image 2.5 Sunbursttext_to_imagegeneration_time112.476080179214 secondsGPT Image 2.5 Sunburst (max), OpenAI — OpenAI; 1024×1024 or nearest supported; 1 image; 3-day median; includes download; Artificial Analysis; retrieved 2026-09-13
GPT Image 2.5 Flaretext_to_imageprice210.7 USD / 1,000 imagesGPT Image 2.5 Flare (max) — OpenAI; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
GPT Image 2.5 Flareimage_editingprice210.7 USD / 1,000 imagesGPT Image 2.5 Flare (max) — OpenAI; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
GPT Image 2.5 Flaretext_to_imagegeneration_time53.0124175548553 secondsGPT Image 2.5 Flare (max), OpenAI — OpenAI; 1024×1024 or nearest supported; 1 image; 3-day median; includes download; Artificial Analysis; retrieved 2026-09-13
GPT Image 2text_to_imageprice211.0 USD / 1,000 imagesGPT Image 2 (high) — OpenAI; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
GPT Image 2image_editingprice211.0 USD / 1,000 imagesGPT Image 2 (high) — OpenAI; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Gemini 3.1 Flash Imagetext_to_imageprice67.0 USD / 1,000 imagesNano Banana 2 (Gemini 3.1 Flash Image) — Google; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Gemini 3.1 Flash Imageimage_editingprice67.0 USD / 1,000 imagesNano Banana 2 (Gemini 3.1 Flash Image) — Google; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Doubao Seedream 5.0 Protext_to_imageprice90.0 USD / 1,000 imagesSeedream 5.0 Pro — ByteDance Seed; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Doubao Seedream 5.0 Proimage_editingprice90.0 USD / 1,000 imagesSeedream 5.0 Pro — ByteDance Seed; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Qwen Image 3.0 Protext_to_imageprice40.0 USD / 1,000 imagesQwen-Image-3.0-Pro — Alibaba; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Qwen Image 3.0 Proimage_editingprice40.0 USD / 1,000 imagesQwen-Image-3.0-Pro — Alibaba; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
FLUX.2 Protext_to_imageprice30.0 USD / 1,000 imagesFLUX.2 [pro] — Black Forest Labs; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
FLUX.2 Proimage_editingprice45.0 USD / 1,000 imagesFLUX.2 [pro] — Black Forest Labs; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
FLUX.2 Protext_to_imagegeneration_time11.2778015136719 secondsFLUX.2 [pro], Black Forest Labs — Black Forest Labs; 1024×1024 or nearest supported; 1 image; 3-day median; includes download; Artificial Analysis; retrieved 2026-09-13
Doubao Seedream 4.5text_to_imageprice40.0 USD / 1,000 imagesSeedream 4.5 — ByteDance Seed; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Doubao Seedream 4.5image_editingprice40.0 USD / 1,000 imagesSeedream 4.5 — ByteDance Seed; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Doubao Seedream 5.0 Litetext_to_imageprice35.0 USD / 1,000 imagesSeedream 5.0 Lite — ByteDance Seed; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Doubao Seedream 5.0 Liteimage_editingprice35.0 USD / 1,000 imagesSeedream 5.0 Lite — ByteDance Seed; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Wan 2.7 Image Protext_to_imageprice64.0 USD / 1,000 imagesWan 2.7 Pro — Alibaba; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Wan 2.7 Imagetext_to_imageprice26.0 USD / 1,000 imagesWan 2.7 — Alibaba; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Wan 2.7 Imageimage_editingprice26.0 USD / 1,000 imagesWan 2.7 — Alibaba; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Step Image Edit 2text_to_imageprice3.0 USD / 1,000 imagesStep Image Edit 2 — StepFun; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Step Image Edit 2image_editingprice3.0 USD / 1,000 imagesStep Image Edit 2 — StepFun; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13
Agnes Image 2.0 Flashimage_editingprice3.0 USD / 1,000 imagesAgnes Image 2.0 Flash — Sapiens AI; 1024×1024; creator API; default settings (AA pricing footnote); Artificial Analysis; retrieved 2026-09-13

Video generation models

Quality coverage: 15/18 models, 4 boards from 2 evaluation operators. Public preference evidence, not comprehensive capability certification or a ChinaAPI endpoint test. Tasks and audio pools are separate; scores across columns are not interchangeable. Overlapping intervals do not establish a significant lead.

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.

Video generation models, public benchmark results as of 2026-09-22
#ModelVendorOriginReleasedOn ChinaAPILMArena Text-to-Video ArenaLMArena Image-to-Video ArenaArtificial Analysis Text To Video · With AudioArtificial Analysis Image To Video · With AudioNotes
1Gemini Omni Flash (video)GoogleInternational—no1511 ±10 — gemini-omni-flash; text_to_video; 95% CI; samples not captured; retrieved 2026-09-121462 ±6 — gemini-omni-flash; image_to_video; 95% CI; samples not captured; retrieved 2026-09-121238 -6/6 — Gemini Omni Flash; text_to_video_audio; 95% CI; samples 16372; retrieved 2026-09-131181 -7/7 — Gemini Omni Flash; image_to_video_audio; 95% CI; samples 11592; retrieved 2026-09-13
2Wan 3.0 VideoAlibabaChina2026-08-13yes1494 ±19 — wan3.0; text_to_video; 95% CI; samples 1167; retrieved 2026-09-121481 ±16 — wan3.0; image_to_video; 95% CI; samples 1401; retrieved 2026-09-121242 -9/9 — Wan 3.0; text_to_video_audio; 95% CI; samples 5746; retrieved 2026-09-131178 -8/8 — Wan 3.0; image_to_video_audio; 95% CI; samples 7876; retrieved 2026-09-13Scores quote the arena's wan3.0 rows, linked to Alibaba's Wan3.0 release. Read on 2026-09-12: 1,167 text-to-video and 1,401 image-to-video votes; the reported intervals are wide. These scores are not attributed to the separate Prime SKU.
3Doubao Seedance 2.5ByteDanceChina2026-06-28yes1482 ±12 — dreamina-seedance-2.5-720p; text_to_video; 95% CI; samples 4066; retrieved 2026-09-121478 ±10 — dreamina-seedance-2.5-720p; image_to_video; 95% CI; samples 6725; retrieved 2026-09-12——The arena entry is the 720p Dreamina build of Seedance 2.5. Read on 2026-09-12: 4,066 text-to-video and 6,725 image-to-video votes; the reported intervals remain wide.
4Doubao Seedance 2.0ByteDanceChina2026-01-28yes1479 ±8 — dreamina-seedance-2.0-720p; text_to_video; 95% CI; samples not captured; retrieved 2026-09-121477 ±8 — dreamina-seedance-2.0-720p; image_to_video; 95% CI; samples not captured; retrieved 2026-09-121220 -6/6 — Dreamina Seedance 2.0 720p; text_to_video_audio; 95% CI; samples 22625; retrieved 2026-09-131196 -6/6 — Dreamina Seedance 2.0 720p; image_to_video_audio; 95% CI; samples 17072; retrieved 2026-09-13The arena entry is the 720p Dreamina build of Seedance 2.0.
5MiniMax H3MiniMaxChina2026-07-31yes1462 ±10 — minimax-h3; text_to_video; 95% CI; samples not captured; retrieved 2026-09-121497 ±6 — minimax-h3; image_to_video; 95% CI; samples not captured; retrieved 2026-09-121225 -7/7 — MiniMax H3; text_to_video_audio; 95% CI; samples 8854; retrieved 2026-09-131190 -8/8 — MiniMax H3; image_to_video_audio; 95% CI; samples 7461; retrieved 2026-09-13
6Sora 2 ProOpenAIInternational2025-09-30no1367 ±7 — sora-2-pro; text_to_video; 95% CI; samples not captured; retrieved 2026-09-12—1076 -9/9 — Sora 2 Pro; text_to_video_audio; 95% CI; samples 4867; retrieved 2026-09-13—
7Veo 3.1GoogleInternational2025-10-15no1364 ±14 — veo-3.1-audio; text_to_video; 95% CI; samples not captured; retrieved 2026-09-121398 ±11 — veo-3.1-audio; image_to_video; 95% CI; samples not captured; retrieved 2026-09-121090 -6/6 — Veo 3.1; text_to_video_audio; 95% CI; samples 8976; retrieved 2026-09-131088 -7/7 — Veo 3.1; image_to_video_audio; 95% CI; samples 8617; retrieved 2026-09-13
8Wan 2.7 T2VAlibabaChina2026-04-03yes1341 ±8 — wan2.7-t2v; text_to_video; 95% CI; samples not captured; retrieved 2026-09-12—1104 -7/7 — Wan 2.7; text_to_video_audio; 95% CI; samples 5986; retrieved 2026-09-13—
9MiniMax Hailuo 2.3MiniMaxChina2025-10-28yes1205 ±6 — hailuo-2.3; text_to_video; 95% CI; samples not captured; retrieved 2026-09-121261 ±5 — hailuo-2.3; image_to_video; 95% CI; samples not captured; retrieved 2026-09-12——
10MiniMax Hailuo 02MiniMaxChina—yes1198 ±13 — hailuo-02-pro; text_to_video; 95% CI; samples not captured; retrieved 2026-09-121228 ±11 — hailuo-02-pro; image_to_video; 95% CI; samples not captured; retrieved 2026-09-12——Scores are the arena's hailuo-02-pro rows; the arena also carries standard and fast builds of the same family.
11Kling v3KuaishouChina2026-02-05yes—1356 ±6 — kling-v3-pro; image_to_video; 95% CI; samples not captured; retrieved 2026-09-121108 -5/5 — Kling 3.0 1080p (Pro); text_to_video_audio; 95% CI; samples 20826; retrieved 2026-09-131076 -6/6 — Kling 3.0 1080p (Pro); image_to_video_audio; 95% CI; samples 16317; retrieved 2026-09-13The arena entry is the pro build; the text-to-video board has no v3 row at the time of this snapshot.
12Kling v3 OmniKuaishouChina2026-02-05yes——1090 -6/6 — Kling 3.0 Omni 1080p (Pro); text_to_video_audio; 95% CI; samples 8871; retrieved 2026-09-131066 -7/7 — Kling 3.0 Omni 1080p (Pro); image_to_video_audio; 95% CI; samples 7665; retrieved 2026-09-13No row on the LMArena video boards at the time of this snapshot. Artificial Analysis evidence is now quoted separately for the identified task and exact entry.
13Wan 2.7 I2VAlibabaChina2026-04-03yes—1428 ±5 — wan2.7-i2v; image_to_video; 95% CI; samples not captured; retrieved 2026-09-12—1088 -7/7 — Wan 2.7; image_to_video_audio; 95% CI; samples 5703; retrieved 2026-09-13
14HappyHorse 1.1 I2VAlibabaChina—yes———1109 -7/7 — HappyHorse-1.1; image_to_video_audio; 95% CI; samples 13810; retrieved 2026-09-13No HappyHorse 1.1 row on the LMArena video boards as read on 2026-09-12. The arena carries the previous generation, happyhorse-1.0; that build's scores are deliberately not reused here. Artificial Analysis evidence is now quoted separately for the identified task and exact entry.
15HappyHorse 1.1 T2VAlibabaChina—yes——1146 -6/6 — HappyHorse-1.1; text_to_video_audio; 95% CI; samples 18445; retrieved 2026-09-13—No HappyHorse 1.1 row on the LMArena video boards as read on 2026-09-12. The arena carries the previous generation, happyhorse-1.0; that build's scores are deliberately not reused here. Artificial Analysis evidence is now quoted separately for the identified task and exact entry.
16Kling 3.0 TurboKuaishouChina—yes————No row on the LMArena video boards at the time of this snapshot.
17MiniMax Hailuo 2.3 FastMiniMaxChina2025-10-28yes————No row on the LMArena video boards at the time of this snapshot.
18Wan 3.0 Video PrimeAlibabaChina—yes————No separately identified Wan 3.0 Prime row on either Arena video board as read on 2026-09-12. The wan3.0 row does not identify Prime, so its scores are not reused here.

Third-party price and generation-time references

Not ChinaAPI prices or endpoint measurements. These references never affect quality coverage or ranking. Compare only matching tasks and configurations. Generation time is a trailing-three-day median of successful requests including download, not retry overhead.

ModelTaskMetricValueEvidence
Gemini Omni Flash (video)text_to_video_audioprice6.00 USD / video minuteGemini Omni Flash — Google; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Gemini Omni Flash (video)image_to_video_audioprice6.00 USD / video minuteGemini Omni Flash — Google; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Wan 3.0 Videotext_to_video_audioprice12.00 USD / video minuteWan 3.0 — Alibaba; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Wan 3.0 Videoimage_to_video_audioprice12.00 USD / video minuteWan 3.0 — Alibaba; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Doubao Seedance 2.0text_to_video_audioprice9.07 USD / video minuteDreamina Seedance 2.0 720p — ByteDance Seed; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Doubao Seedance 2.0image_to_video_audioprice9.07 USD / video minuteDreamina Seedance 2.0 720p — ByteDance Seed; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
MiniMax H3text_to_video_audioprice7.80 USD / video minuteMiniMax H3 — MiniMax; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
MiniMax H3image_to_video_audioprice7.80 USD / video minuteMiniMax H3 — MiniMax; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Sora 2 Protext_to_video_audioprice30.00 USD / video minuteSora 2 Pro — OpenAI; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Veo 3.1text_to_video_audioprice24.00 USD / video minuteVeo 3.1 — Google; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Veo 3.1image_to_video_audioprice24.00 USD / video minuteVeo 3.1 — Google; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Wan 2.7 T2Vtext_to_video_audioprice9.00 USD / video minuteWan 2.7 — Alibaba; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Kling v3text_to_video_audioprice20.16 USD / video minuteKling 3.0 1080p (Pro) — KlingAI; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Kling v3image_to_video_audioprice20.16 USD / video minuteKling 3.0 1080p (Pro) — KlingAI; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Kling v3text_to_video_silentgeneration_time99.7195221020002 secondsKling 3.0 1080p (Pro), fal.ai — fal.ai; 720p / 5s panel target; nearest supported settings (Kling Pro: 1080p); no audio; 3-day median; includes download; Artificial Analysis; retrieved 2026-09-13
Kling v3 Omnitext_to_video_audioprice16.80 USD / video minuteKling 3.0 Omni 1080p (Pro) — KlingAI; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Kling v3 Omniimage_to_video_audioprice16.80 USD / video minuteKling 3.0 Omni 1080p (Pro) — KlingAI; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
Wan 2.7 I2Vimage_to_video_audioprice9.00 USD / video minuteWan 2.7 — Alibaba; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
HappyHorse 1.1 I2Vimage_to_video_audioprice9.90 USD / video minuteHappyHorse-1.1 — Alibaba-ATH; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
HappyHorse 1.1 T2Vtext_to_video_audioprice9.90 USD / video minuteHappyHorse-1.1 — Alibaba-ATH; 1080p; creator API; default settings (AA pricing footnote, not the quality row resolution); Artificial Analysis; retrieved 2026-09-13
HappyHorse 1.1 T2Vtext_to_video_silentgeneration_time107.870128135 secondsHappyHorse-1.1, Alibaba Cloud (T2V) — Alibaba Cloud (T2V); 720p / 5s panel target; nearest supported settings (Kling Pro: 1080p); no audio; 3-day median; includes download; Artificial Analysis; retrieved 2026-09-13

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 share of calls ChinaAPI served over the last 7 days
#ModelVendorShare of callsTrend vs the previous 7 days
1glm-5.3-flash智谱5.9%up
2deepseek-flashDeepSeek4.6%down
3hy3Tencent4.4%up
4claude-opus-4-8Anthropic2.9%up
5gpt-5.6-solOpenAI2.5%up
6MiniMax-M3MiniMax1.9%flat
7agnes-3.0-flashAgnes AI1.7%up
8kimi-k2.6Moonshot1.7%flat
9claude-opus-5Anthropic1.5%up
10gpt-5.6-lunaOpenAI1.5%up
11claude-sonnet-5Anthropic1.4%flat
12deepseek-v4-proDeepSeek1.4%down
13gpt-5.6-terraOpenAI1.3%up
14claude-opus-4-7Anthropic1.3%flat
15claude-fable-5Anthropic1.3%flat
16claude-sonnet-4-6Anthropic1.2%flat
17claude-opus-4-6Anthropic1.2%flat
18claude-haiku-4-5Anthropic1.2%up
19step-audio-2StepFun1.1%flat
20agnes-2.5-flashAgnes AI1.0%down
All other models58.9%

How those calls arrived: 9.9% over agent protocols (Claude Messages and the OpenAI Responses API), 89.1% over chat protocols (OpenAI chat completions and Gemini generateContent), 1.0% over everything else (audio, images, embeddings, rerank and asynchronous tasks).

Share of calls served, the last 30 days

Model share of calls ChinaAPI served over the last 30 days
#ModelVendorShare of callsTrend vs the previous 30 days
1agnes-2.5-flashAgnes AI4.6%up
2deepseek-flashDeepSeek4.4%up
3deepseek-v4-proDeepSeek4.3%down
4glm-5.3-flash智谱3.3%up
5gpt-5.6-solOpenAI2.1%flat
6kimi-k2.7-codeMoonshot2.0%up
7hy3Tencent1.8%up
8claude-opus-4-8Anthropic1.7%flat
9gpt-5.6-lunaOpenAI1.6%flat
10kimi-k2.6Moonshot1.5%up
11claude-opus-5Anthropic1.4%flat
12claude-sonnet-5Anthropic1.3%down
13MiniMax-M3MiniMax1.3%up
14kimi-k3Moonshot1.2%down
15claude-opus-4-7Anthropic1.2%up
16claude-sonnet-4-6Anthropic1.2%up
17gpt-5.6-terraOpenAI1.2%flat
18claude-opus-4-6Anthropic1.2%up
19claude-fable-5Anthropic1.1%flat
20glm-5.2智谱1.1%down
All other models60.3%

How those calls arrived: 4.5% over agent protocols (Claude Messages and the OpenAI Responses API), 93.6% over chat protocols (OpenAI chat completions and Gemini generateContent), 1.9% over everything else (audio, images, embeddings, rerank and asynchronous tasks).