{ "meta": { "attribution": "Anchor Terminal (https://www.anchorterminal.com)", "attributionRequired": true, "credit": "CC BY 4.0, credit Anchor Terminal and link the page or API address you took it from (https://www.anchorterminal.com/terms/#data)", "docs": "https://www.anchorterminal.com/docs/", "generatedAt": "2026-11-11", "license": "CC-BY-4.0", "method": "https://www.anchorterminal.com/benchmark/", "methodology": "0.7", "openapi": "https://www.anchorterminal.com/openapi.json", "preview": false, "run": "2026-10-01", "runLabel": "October 2026 research run", "terms": "https://www.anchorterminal.com/terms/#data" }, "tool": { "slug": "mimo-v2-6-flash-rl", "name": "MiMo-V2.6-Flash-RL", "vendor": "Xiaomi (MiMo team)", "vendorUrl": "https://mimo.xiaomi.com", "kind": "weights", "category": "open-weights", "summary": "MiMo-V2.6-Flash-RL is Xiaomi's open-weight agent model, a mixture of experts of 309B parameters with 15B active, taking text, images, video and audio with a 1M context. MIT. The weights are 178 GB at 4 bits, so it needs a server.", "url": "https://www.anchorterminal.com/tools/mimo-v2-6-flash-rl", "markdownUrl": "https://www.anchorterminal.com/tools/mimo-v2-6-flash-rl.md", "slimMarkdownUrl": "https://www.anchorterminal.com/tools/mimo-v2-6-flash-rl.min.md", "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/mimo-v2-6-flash-rl.json", "dns": "mimo-v2-6-flash-rl.grade.anchorterminal.com", "repo": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL", "license": "MIT (metadata only, no LICENSE file)", "transports": [], "packages": [ { "registry": "huggingface", "name": "XiaomiMiMo/MiMo-V2.6-Flash-RL" } ], "auth": "none", "authNotes": "The weights download from Hugging Face with no account, token or terms to accept. The repository is not gated (checked 2026-11-10).", "pricing": "free", "pricingNotes": "Free to download and use under MIT, per the card's metadata. Running it yourself needs a multi-GPU server: the files are 177.7 GB with most weights at 4 bits. Xiaomi's API platform and Novita serve it for those without the hardware (checked 2026-11-10).", "priceSummary": "Free", "where": "local", "x402": { "level": "no", "evidence": "Weights you download and run yourself, so there is nothing to pay per call and no payment protocol (checked 2026-11-10).", "endpoints": [] }, "toolCount": null, "popularity": { "githubStars": null, "npmWeekly": null, "pypiWeekly": null, "asOf": "2026-11-10" }, "docsUrl": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL", "capabilities": [ "model.open-weights" ], "tags": [ "open-weights", "model", "moe", "multimodal", "vision", "audio", "video", "reasoning", "tool-calling", "agentic", "long-context", "mit", "safetensors", "fp4", "custom-code", "self-hosted" ], "lastRelease": "2026-10-28", "graded": true, "anchor": { "graded": true, "score": 61, "grade": "C", "agentReady": false, "rank": 0, "ranked": false, "notRankedWhy": "An open-weight model, graded by the open-weight checklist and ranked among open-weight models, not against tools", "rankOf": 1518, "categoryRank": 59, "methodology": "0.7", "run": "2026-10-01", "scores": { "interface": 85, "reach": 65, "risk": 45, "upkeep": 59 }, "pending": [ "performance", "tasks" ], "breakdown": [ { "key": "capability", "name": "Capability", "weight": 35, "effectiveWeight": 0, "pending": true, "points": 0, "reason": "Pending for this listing. Nobody independent has published results for it in at least two areas yet, so its weight is shared across the other four and the total is capped at 61 until they do or our own runs start." }, { "key": "reach", "name": "Reach", "weight": 25, "effectiveWeight": 38.46, "score": 65, "points": 25, "reason": "Memory 5, the smallest build at 4 bits or better is ggml-org's MXFP4 GGUF, 168.93 GB with its projector, and 168.93 GB × 1.2 = 202.7 GB, within 256 GB. Xiaomi's own files are 177.74 GB. ggml-org's Q2_K is below 4 bits and isn't read. Runtimes 20, Released vLLM 0.33.0 and SGLang 0.5.23 have the mimo_v2 model and the mimo parsers, Transformers 5.22.2 has a mimo_v2_flash model, and llama.cpp b12242 has the architecture, with a GGUF from its own organisation. Builds 10, one precision from Xiaomi, the 4-bit files it releases. Terms 15, MIT in the card's metadata, an OSI licence. Download 10, the Hub API reports gated false on 10 November 2026, so the files download with no account and nothing to accept. Hosted 5, the card says it is on Xiaomi's MiMo Open Platform API and on OpenRouter, and Novita is live for it on Hugging Face. 5 + 20 + 10 + 15 + 10 + 5 = 65.", "checked": "2026-11-10" }, { "key": "risk", "name": "Risk", "weight": 20, "effectiveWeight": 30.77, "score": 45, "points": 13.85, "reason": "File safety 5, the weights are safetensors, but the dflash folder's mask_embedding.pt is a pickle beside them (15). Provenance 15, the files sit in the XiaomiMiMo organisation, which mimo.xiaomi.com links to per the Distill listing (10). Safety evaluation 0, no safety results on the card. Agent guidance 5, General cautions only. Licence risk 15, MIT has no use policy, threshold or change clause. Security programme 5, Xiaomi runs a company security centre for reporting flaws, as read for the Distill listing (5). 5 + 15 + 0 + 5 + 15 + 5 = 45.", "checked": "2026-11-10" }, { "key": "interface", "name": "Agent interface", "weight": 10, "effectiveWeight": 15.38, "score": 85, "points": 13.08, "reason": "Chat template 20, chat_template.jinja in the repository (3,868 characters), with a copy in tokenizer_config.json. Tool-calling format 25, the card names the parser to use in two runtimes: the mimo tool-call parser in SGLang and in vLLM, with auto tool choice, and both ship it in a release. Trained tool calling 20, the card describes reinforcement learning on coding and general agent tasks across several harnesses, and the template renders every call in a turn. Structured output 0, Nothing on the card about JSON or schema-constrained output. Reasoning control 10, the template has an enable_thinking switch and keeps reasoning_content, and the card names the mimo reasoning parser for SGLang and vLLM, which returns the reasoning apart from the answer. Context 10, 1M tokens per the card's model summary, and 1,048,576 positions in config.json. 20 + 25 + 20 + 0 + 10 + 10 = 85.", "checked": "2026-11-10" }, { "key": "upkeep", "name": "Documentation and upkeep", "weight": 10, "effectiveWeight": 15.38, "score": 59, "points": 9.08, "reason": "Model card 14, Intended use, the architecture, the parameters, the context and the languages (in the metadata) are on the card. Training data 5, A description of the training method with no data sources, sizes or cut-off. Last change 20, the last commits, on 28 October 2026, updated the card, 13 days before this check. Responsiveness 5, Of the 13 most recent discussions that aren't pull requests (21 September to 6 November 2026), 3 have a reply from a XiaomiMiMo member, which is a few. Version policy 5, MiMo-V2.5 stays in its own repositories and V2.6 came out in new ones, with no dated note on the older cards that we read. File health 10, Config, template and card agree on the architecture and the 1M context, and two early faults (invalid JSON in the draft model's config, preprocessor values) were fixed in commits that say so. 14 + 5 + 20 + 5 + 5 + 10 = 59.", "checked": "2026-11-10" } ], "assessment": { "date": "2026-11-10", "basis": "public evidence", "confidence": "low", "notes": { "capability": "Pending for this listing. Independent results were found in 0 areas (none), and Capability needs 2. Its weight is shared across the other four categories and the total is capped at 61 until there are more, or our own runs start.", "interface": "Chat template 20, chat_template.jinja in the repository (3,868 characters), with a copy in tokenizer_config.json. Tool-calling format 25, the card names the parser to use in two runtimes: the mimo tool-call parser in SGLang and in vLLM, with auto tool choice, and both ship it in a release. Trained tool calling 20, the card describes reinforcement learning on coding and general agent tasks across several harnesses, and the template renders every call in a turn. Structured output 0, Nothing on the card about JSON or schema-constrained output. Reasoning control 10, the template has an enable_thinking switch and keeps reasoning_content, and the card names the mimo reasoning parser for SGLang and vLLM, which returns the reasoning apart from the answer. Context 10, 1M tokens per the card's model summary, and 1,048,576 positions in config.json. 20 + 25 + 20 + 0 + 10 + 10 = 85.", "reach": "Memory 5, the smallest build at 4 bits or better is ggml-org's MXFP4 GGUF, 168.93 GB with its projector, and 168.93 GB × 1.2 = 202.7 GB, within 256 GB. Xiaomi's own files are 177.74 GB. ggml-org's Q2_K is below 4 bits and isn't read. Runtimes 20, Released vLLM 0.33.0 and SGLang 0.5.23 have the mimo_v2 model and the mimo parsers, Transformers 5.22.2 has a mimo_v2_flash model, and llama.cpp b12242 has the architecture, with a GGUF from its own organisation. Builds 10, one precision from Xiaomi, the 4-bit files it releases. Terms 15, MIT in the card's metadata, an OSI licence. Download 10, the Hub API reports gated false on 10 November 2026, so the files download with no account and nothing to accept. Hosted 5, the card says it is on Xiaomi's MiMo Open Platform API and on OpenRouter, and Novita is live for it on Hugging Face. 5 + 20 + 10 + 15 + 10 + 5 = 65.", "risk": "File safety 5, the weights are safetensors, but the dflash folder's mask_embedding.pt is a pickle beside them (15). Provenance 15, the files sit in the XiaomiMiMo organisation, which mimo.xiaomi.com links to per the Distill listing (10). Safety evaluation 0, no safety results on the card. Agent guidance 5, General cautions only. Licence risk 15, MIT has no use policy, threshold or change clause. Security programme 5, Xiaomi runs a company security centre for reporting flaws, as read for the Distill listing (5). 5 + 15 + 0 + 5 + 15 + 5 = 45.", "upkeep": "Model card 14, Intended use, the architecture, the parameters, the context and the languages (in the metadata) are on the card. Training data 5, A description of the training method with no data sources, sizes or cut-off. Last change 20, the last commits, on 28 October 2026, updated the card, 13 days before this check. Responsiveness 5, Of the 13 most recent discussions that aren't pull requests (21 September to 6 November 2026), 3 have a reply from a XiaomiMiMo member, which is a few. Version policy 5, MiMo-V2.5 stays in its own repositories and V2.6 came out in new ones, with no dated note on the older cards that we read. File health 10, Config, template and card agree on the architecture and the 1M context, and two early faults (invalid JSON in the draft model's config, preprocessor values) were fixed in commits that say so. 14 + 5 + 20 + 5 + 5 + 10 = 59." }, "sources": [ { "what": "model card", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL", "seen": "2026-11-10" }, { "what": "gating, dates, parameter count by precision, downloads, likes, file list and hashes", "url": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Flash-RL?blobs=true", "seen": "2026-11-10" }, { "what": "inference providers, Novita live", "url": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Flash-RL?expand[]=inferenceProviderMapping", "seen": "2026-11-10" }, { "what": "commit history", "url": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Flash-RL/commits/main", "seen": "2026-11-10" }, { "what": "architecture, auto_map and positions", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL/raw/main/config.json", "seen": "2026-11-10" }, { "what": "chat template", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL/raw/main/chat_template.jinja", "seen": "2026-11-10" }, { "what": "discussions, replies counted by script from authors, dates and member flags", "url": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Flash-RL/discussions", "seen": "2026-11-10" }, { "what": "ggml-org's GGUF repository: files and sizes", "url": "https://huggingface.co/api/models/ggml-org/MiMo-V2.6-Flash-RL-GGUF?blobs=true", "seen": "2026-11-10" }, { "what": "vLLM v0.33.0, mimo_v2.py and the mimo parsers", "url": "https://raw.githubusercontent.com/vllm-project/vllm/v0.33.0/vllm/model_executor/models/registry.py", "seen": "2026-11-10" }, { "what": "SGLang v0.5.23, srt/models/mimo_v2.py and the parser names", "url": "https://raw.githubusercontent.com/sgl-project/sglang/v0.5.23/python/sglang/srt/models/mimo_v2.py", "seen": "2026-11-10" }, { "what": "Transformers v5.22.2 mimo_v2_flash", "url": "https://github.com/huggingface/transformers/tree/v5.22.2/src/transformers/models/mimo_v2_flash", "seen": "2026-11-10" }, { "what": "llama.cpp b12242, the mimo architecture (shallow clone)", "url": "https://raw.githubusercontent.com/ggml-org/llama.cpp/b12242/src/llama-arch.cpp", "seen": "2026-11-10" }, { "what": "our MiMo-V2.6 Distill listing, for company facts", "url": "https://www.anchorterminal.com/tools/mimo-v2-6-distill-qwen-9b", "seen": "2026-11-10" } ], "openQuestions": [ "File safety is 5: 15 for a pickle in the optional dflash folder, less 10 for trust-remote-code in the card's commands. Someone who skips that folder and uses Transformers natively runs neither", "It was on the gap list as fitting a 128 GB machine. By the calculator it doesn't: its smallest 4-bit build needs about 203 GB", "Reasoning control is 10 on the template's switch and the named parser. The card shows no request that turns thinking off", "MiMo-V2.6-Pro-RL, the larger model the card links, isn't listed", "Capability is pending because no independent results were looked up on the day", "unchecked on 10 November: the technical report, the repository's Python files, SGLang's cookbook and vLLM's recipe, Xiaomi's API platform, OpenRouter, Ollama's library and LM Studio's catalogue" ] }, "negative": 0, "verdict": "A large agent model with an open licence: text, images, video and audio, a 1M context, tool calling it was trained for, and code in four released runtimes. It's a server model, 178 GB at 4 bits, and the card's commands run the repository's code. No safety results, little said about data, and few questions answered.", "bestFor": "Teams with a GPU server, or a provider account, that want an open agent model with a very long context", "strengths": [ "MIT, ungated, with hosted options", "Four released runtimes carry it, with tool and reasoning parsers in vLLM and SGLang", "A 1M context and four input types", "Trained with reinforcement learning across agent harnesses", "A GGUF from the llama.cpp project's own organisation" ], "weaknesses": [ "178 GB at 4 bits, so it needs about 200 GB of memory", "The card's serve commands pass trust-remote-code", "A pickle file sits in the draft model's folder", "No safety results, limitations or data sources on the card", "Xiaomi replied in a few of the discussions only" ], "agentNotes": [ "Plan for a provider unless you have a multi-GPU server: Xiaomi's API platform, OpenRouter or Novita", "On SGLang or vLLM pass `--tool-call-parser mimo --reasoning-parser mimo`", "Use temperature 1.0 and top_p 0.95, as the card recommends", "The card's commands pass trust-remote-code. Read the repository's three Python files first, or use Transformers 5.22.2, which has the model built in", "Skip the dflash folder unless you need speculative decoding. Its mask_embedding.pt is a pickle", "The Use Restrictions section asks for human oversight of high-risk actions" ], "metrics": { "kind": "local", "measured": false }, "reviewCount": 0, "avgRating": 0, "history": [ { "basis": "public evidence", "confidence": "low", "grade": "C", "methodology": "0.7", "pending": [ "performance", "tasks" ], "run": "2026-10-01", "runLabel": "October 2026 research run", "score": 61 } ], "editorialScores": { "interface": 85, "reach": 65, "risk": 45, "upkeep": 59 }, "provenanceScore": 58, "checked": { "categories": { "interface": "2026-11-10", "reach": "2026-11-10", "risk": "2026-11-10", "upkeep": "2026-11-10" }, "last": "2026-11-10", "oldest": "2026-11-10", "stale": false, "staleAfterDays": 90 }, "scoreHistory": [ { "date": "2026-11-10", "grade": "C", "kind": "listed", "method": "0.7", "score": 61 } ] }, "connect": { "install": "pip install \"sglang\u003e=0.5.23\"\nsglang serve --trust-remote-code --model-path XiaomiMiMo/MiMo-V2.6-Flash-RL --tp 8 --reasoning-parser mimo --tool-call-parser mimo\n# the card's full command adds data-parallel attention and speculative decoding flags", "http": "# SGLang and vLLM serve an OpenAI-compatible API. The card recommends temperature 1.0 and top_p 0.95,\n# and names the mimo tool-call and reasoning parsers for both" }, "letme": { "capability": "https://letme.dev/model.open-weights", "tool": "https://letme.dev/mimo-v2-6-flash-rl" }, "sameCompany": [ "mimo-v2-6-distill-qwen-9b" ], "weights": { "trainedBy": "Xiaomi (MiMo team)", "params": "309B", "active": "15B", "context": 1048576, "goodFor": "Long agent runs over code, tools, screens and audio on a multi-GPU server, or through a provider.", "runtimes": [ "llama.cpp", "vLLM", "SGLang", "Transformers" ], "builds": [ { "name": "MiMo-V2.6-Flash-RL", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL", "who": "publisher", "precision": "fp4", "sizeGB": 177.74, "memoryGB": 256, "format": "safetensors", "runsOn": [ "vLLM", "SGLang", "Transformers" ], "note": "The only build Xiaomi publishes, with most weights at 4 bits" }, { "name": "ggml-org MiMo-V2.6-Flash-RL-GGUF (MXFP4)", "url": "https://huggingface.co/ggml-org/MiMo-V2.6-Flash-RL-GGUF", "who": "llama.cpp", "precision": "mxfp4", "sizeGB": 168.93, "memoryGB": 256, "format": "gguf", "runsOn": [ "llama.cpp" ], "note": "From ggml-org, the llama.cpp project's organisation. 167.37 GB of MXFP4 weights plus the 1.56 GB Q8_0 projector, from the Hub's model information on 10 November 2026" } ], "run": "sglang serve --trust-remote-code --model-path XiaomiMiMo/MiMo-V2.6-Flash-RL --tp 8 --reasoning-parser mimo --tool-call-parser mimo", "runBuild": "MiMo-V2.6-Flash-RL", "checked": "2026-11-10", "minMemoryGB": 256 }, "family": { "lab": "Xiaomi (MiMo)", "family": "MiMo-V2 (V2, V2.5, V2.6)", "url": "https://www.anchorterminal.com/open-weights/#xiaomi", "others": [ { "name": "MiMo-V2.6-Distill-Qwen-9B", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B", "params": "9B (Hub safetensors count 9.41B)", "listing": "https://www.anchorterminal.com/tools/mimo-v2-6-distill-qwen-9b" }, { "name": "MiMo-V2.6-Pro-MOPD", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-MOPD", "params": "1024.22B (Hub safetensors count; the name states none)" }, { "name": "MiMo-V2.6-Flash-MOPD", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-MOPD", "params": "310.76B (Hub safetensors count; the name states none)" }, { "name": "MiMo-V2.6-Pro-RL", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL", "params": "1024.22B (Hub safetensors count; the name states none)" }, { "name": "MiMo-V2.5-Pro-Base", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5-Pro-Base", "params": "1023.18B (Hub safetensors count; the name states none)" }, { "name": "MiMo-V2.5-Pro", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5-Pro", "params": "1023.18B (Hub safetensors count; the name states none)" }, { "name": "MiMo-V2.5-Base", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5-Base", "params": "310.76B (Hub safetensors count; the name states none)" }, { "name": "MiMo-V2.5", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5", "params": "310.76B (Hub safetensors count; the name states none)" }, { "name": "MiMo-V2.5-ASR", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5-ASR", "params": "7.62B (Hub safetensors count; the name states none)" }, { "name": "MiMo-V2-Flash-Base", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2-Flash-Base", "params": "309.77B (Hub safetensors count; the name states none)" }, { "name": "MiMo-V2-Flash", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2-Flash", "params": "309.77B (Hub safetensors count; the name states none)" } ], "olderCount": 12 }, "notable": [ "76,859 downloads in the last 30 days and 593 likes on Hugging Face on 10 November 2026 (https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Flash-RL)", "309B parameters in total and 15B active per token, per the card. The files hold 310.8B, 302.8B of them packed at 4 bits, 177.7 GB in all (https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Flash-RL?blobs=true)", "One reinforcement learning run across coding, general agents, visual tasks and cybersecurity, mixed in the same batches, per the card (https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL)", "The card's table has it at 67.9 on DeepSWE v1.1 and 87.6 on Terminal Bench 2.1, a few points under MiMo-V2.6 Pro. These are Xiaomi's numbers (https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL)", "The card's Use Restrictions section lists uses it asks you to avoid, military use and unsupervised high-risk actions among them, and says it doesn't amend the licence (https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL)", "ggml-org, the llama.cpp project's organisation, publishes a GGUF of it at MXFP4 (167.4 GB) with projector, draft and MTP files (https://huggingface.co/ggml-org/MiMo-V2.6-Flash-RL-GGUF)", "A Xiaomi member replied in 3 of the 13 most recent discussions (https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Flash-RL/discussions)" ], "area": "models", "details": [ { "label": "Surface graded", "value": "The open weights at huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL, read from the model card, the files and their metadata, the config, the chat template, the commit history, ggml-org's GGUF repository and the runtimes' own code at their latest releases on 10 November 2026. Xiaomi's company facts come from our MiMo-V2.6 Distill listing. The technical report in the repository, the repository's Python files, and the SGLang cookbook and vLLM recipe the card links weren't read. Nothing was downloaded or run" }, { "label": "Model", "value": "310,756,322,688 parameters in the files. The card says a sparse mixture of experts with 309B in total and 15B active, a 681M vision encoder, audio encoders of 308M and 127M, and a five-layer decoder for multi-token prediction. The backbone interleaves sliding-window and global attention" }, { "label": "Files", "value": "67 safetensors files (177.74 GB), with 302.8B parameters stored as packed 4-bit values and the rest in BF16 and FP8, chat_template.jinja, config and tokenizer files, an audio tokenizer, three Python modules, the technical report as a PDF, and a dflash folder with a draft model whose mask_embedding.pt is a pickle. Hugging Face lists a SHA-256 for the large files" }, { "label": "Licence", "value": "MIT in the card's metadata. There is no LICENSE file. The card's Use Restrictions section lists prohibited uses and says it is an informational reminder that doesn't amend the licence or add conditions" }, { "label": "Runtimes", "value": "The card documents SGLang, pointing to SGLang's MiMo cookbook, and vLLM, pointing to vLLM's recipe and a pre-built image, both with trust-remote-code. Released SGLang 0.5.23 and vLLM 0.33.0 have the model and the mimo parsers, Transformers 5.22.2 has a mimo_v2_flash model, and llama.cpp b12242 has the architecture, with ggml-org's GGUF" }, { "label": "Context", "value": "1M tokens per the card, and 1,048,576 positions in config.json" }, { "label": "Tools and reasoning", "value": "The card's SGLang and vLLM commands name the mimo tool-call parser and the mimo reasoning parser, with auto tool choice on vLLM. The chat template takes tools and an enable_thinking switch and keeps reasoning_content. The card itself shows no request with tools" }, { "label": "Hardware", "value": "No memory figure on the card. Its SGLang command uses 8 GPUs with tensor parallelism and its vLLM command 4. The files are 177.7 GB" }, { "label": "Training data", "value": "The card describes the training method at length (mixed reinforcement learning with graders, then distillation from teachers) and names no data sources, sizes or cut-off. The technical report wasn't read" } ], "taskCosts": { "hosted": [ { "openrouterId": "xiaomi/mimo-v2.6-flash", "name": "MiMo-V2.6-Flash", "source": "OpenRouter, $0.14 in and $0.28 out per 1M tokens", "usd": { "bugfix": 0.01, "research": 0.07, "website": 0.03 } } ], "note": "Each figure is the same tokens priced at this model's rates, research report (134 fresh input, 207k cache writes, 10.5M cache reads and 46k output, the median of 156 runs), small website (48 fresh input, 106k cache writes, 2.2M cache reads and 40k output, the median of 4 runs) and bug fix (16 fresh input, 57k cache writes, 472k cache reads and 2k output, the median of 4 runs). We measured the tokens with Claude Opus 5.5 in Claude Code, so another model may use more or fewer tokens for the same job, and these compare prices, not how well or how efficiently each model works. Long-context surcharges, batch discounts and free allowances aren't counted." }, "provenance": { "legalEntity": "Xiaomi (the card and report name Xiaomi MiMo; no LICENSE file names a copyright holder)", "domain": "mimo.xiaomi.com", "domainRegistered": "", "endpointOnVendorDomain": null, "terms": "", "privacy": "", "statusPage": "", "changelog": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL/commits/main", "securityTxt": "none", "checked": "2026-11-10", "notes": [ "Xiaomi's legal entity, domain and security facts are carried from our MiMo-V2.6 Distill listing.", "Nothing is hosted under this listing, so MIT stands in for terms of service and no privacy policy is scored.", "The changelog is the commit history, 13 commits from 21 September to 28 October 2026." ], "score": 58, "checks": [ { "check": "Legal entity named", "value": "Xiaomi (the card and report name Xiaomi MiMo; no LICENSE file names a copyright holder)", "points": 20, "max": 20, "state": "ok" }, { "check": "Domain age", "value": "mimo.xiaomi.com, no registry record we could read", "points": 0, "max": 15, "state": "no" }, { "check": "Endpoint on the vendor's domain", "value": "no hosted endpoint", "points": 0, "max": 0, "state": "na" }, { "check": "Terms of service", "value": "nothing hosted, so the MIT (metadata only, no LICENSE file) licence stands in", "points": 10, "max": 10, "state": "ok" }, { "check": "Privacy policy", "value": "nothing hosted, not scored", "points": 0, "max": 0, "state": "na" }, { "check": "Status page", "value": "not found", "points": 0, "max": 10, "state": "no" }, { "check": "Changelog", "value": "published", "points": 10, "max": 10, "state": "ok" }, { "check": "security.txt", "value": "not found", "points": 0, "max": 10, "state": "no" }, { "check": "People named", "value": "3 named with their role, none with a public profile", "points": 7, "max": 10, "state": "part" }, { "check": "Support route", "value": "a community", "points": 5, "max": 5, "state": "ok" } ] }, "humans": { "slug": "mimo-v2-6-flash-rl", "checked": "2026-11-10", "company": "Xiaomi (MiMo team)", "people": [ { "name": "Lei Jun", "role": "Founder, chairman and CEO, Xiaomi", "founder": true, "photo": "/assets/img/people/xiaomi-mimo-team-lei-jun.webp", "photoSource": "https://www.mi.com/global/about/founder/", "bio": "Lei Jun founded Xiaomi and is its chairman and chief executive. He joined Kingsoft in 1992, held several senior roles there and has chaired its board since July 2011. In 2000 he founded the online retailer joyo.com, which Amazon acquired in 2004, and he has made angel investments in companies including JOYY and UCWeb.", "bioSources": [ "https://www.mi.com/global/about/founder/" ], "page": "/people/lei-jun/" }, { "name": "Lin Bin", "role": "Co-founder, vice chairman and executive director, Xiaomi", "founder": true, "photo": "/assets/img/people/xiaomi-mimo-team-lin-bin.webp", "photoSource": "https://www.mi.com/global/about/founder/", "bio": "Lin Bin co-founded Xiaomi with Lei Jun in 2010 and was its president until 2019, when he became vice chairman. Before Xiaomi he was an engineering director at Google from 2006 to 2010, and from 1995 to 2006 he worked at Microsoft in roles from software design engineer to engineering director.", "bioSources": [ "https://www.mi.com/global/about/founder/" ], "page": "/people/lin-bin/" }, { "name": "Liu De", "role": "Co-founder, executive director and senior vice president, Xiaomi", "founder": true, "photo": "/assets/img/people/xiaomi-mimo-team-liu-de.webp", "photoSource": "https://www.mi.com/global/about/founder/", "bio": "Liu De co-founded Xiaomi in 2010 and is an executive director, a senior vice president and head of its Leadership Management Department. His early work there covered the smartphone supply chain and industrial design. Before Xiaomi he founded the design firms Rethink Concept and Beijing Xinfengrui Industrial Design.", "bioSources": [ "https://www.mi.com/global/about/founder/" ], "page": "/people/liu-de/" } ], "peopleNote": "www.mi.com and ir.mi.com answered 403 to their robots.txt on 28 October 2026, so we fetched nothing from them. Xiaomi's Hong Kong exchange announcement of 18 August 2026, signed by Lei Jun as chairman, gives the board as Lei Jun (chairman and executive director), Lin Bin (vice chairman and executive director) and Liu De (executive director), so all three are still in place. The model card credits the Xiaomi MiMo Team with no named lead, and mimo.xiaomi.com names nobody either. Roster may be incomplete.", "companySocial": { "x": "https://x.com/Xiaomi" }, "support": { "community": [ { "label": "Hugging Face discussions", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL/discussions", "kind": "forum" } ] }, "sources": [ "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL", "https://huggingface.co/XiaomiMiMo", "https://mimo.xiaomi.com", "https://www.mi.com/global/about/founder/" ], "ownership": { "status": "public", "exchange": "HKEX", "ticker": "1810", "checked": "2026-10-28", "sources": [ "https://www1.hkexnews.hk/listedco/listconews/sehk/2026/0818/2026081801086.pdf" ] } }, "pageJsonUrl": "https://www.anchorterminal.com/tools/mimo-v2-6-flash-rl.json" } }