{ "data": { "category": { "area": "models", "capabilities": [ "model.open-weights" ], "description": "Model weights you download and run on your own hardware, under a licence that says what you may do with them. Each is graded on the files, apart from any hosted API that serves the same model. Compared on the licence, the runtimes that load it, tool calling, context, the memory it needs and what the vendor publishes about training data and evaluations.", "json": "https://www.anchorterminal.com/categories/open-weights.json", "name": "Open-weight models", "slug": "open-weights", "test": "In this run, public evidence against the open-weight reading of the checklist, from the model card, the files, the licence and the runtimes' own code. When our local runs start, each model that fits in 128 GB is served on one machine through llama.cpp and vLLM at the largest precision that fits. We measure tokens per second and time to first token, then run the agent task suite with the model driving.", "title": "Open-weight models you download and run yourself", "toolCount": 3, "tools": [ "kolibri-1", "qwen3-8-27b", "deepseek-v4-1-flash" ], "url": "https://www.anchorterminal.com/categories/open-weights" }, "tools": [ { "slug": "kolibri-1", "name": "Kolibri-1", "vendor": "Aleph Alpha", "vendorUrl": "https://aleph-alpha.com", "kind": "weights", "category": "open-weights", "summary": "Kolibri-1 is Aleph Alpha's open-weight mixture-of-experts reasoning model for German and English, 78B parameters with 3.46B active per token, published on Hugging Face in FP8 and BF16 under Apache 2.0 and served with vLLM.", "url": "https://www.anchorterminal.com/tools/kolibri-1", "markdownUrl": "https://www.anchorterminal.com/tools/kolibri-1.md", "slimMarkdownUrl": "https://www.anchorterminal.com/tools/kolibri-1.min.md", "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/kolibri-1.json", "repo": "https://huggingface.co/Aleph-Alpha/Kolibri-1", "license": "Apache-2.0", "transports": [], "packages": [ { "registry": "huggingface", "name": "Aleph-Alpha/Kolibri-1" }, { "registry": "huggingface", "name": "Aleph-Alpha/Kolibri-1-BF16" }, { "registry": "pypi", "name": "aleph-alpha-inference" } ], "auth": "none", "authNotes": "The weights download from Hugging Face with no account, token or terms to accept. Neither repository is gated (checked 2026-10-20).", "pricing": "free", "pricingNotes": "Free to download and use, commercial use included, under Apache 2.0. Aleph Alpha sells no per-call access to these weights. Running them costs the hardware, about 78 GB of accelerator memory for the FP8 weights per the model card (checked 2026-10-20).", "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 2026-10-20).", "endpoints": [] }, "toolCount": null, "popularity": { "githubStars": null, "npmWeekly": null, "pypiWeekly": null, "asOf": "2026-10-20" }, "docsUrl": "https://huggingface.co/Aleph-Alpha/Kolibri-1", "capabilities": [ "model.open-weights" ], "tags": [ "open-weights", "model", "moe", "reasoning", "tool-calling", "german", "apache-2.0", "safetensors", "fp8", "self-hosted", "new" ], "lastRelease": "2026-10-19", "graded": true, "anchor": { "graded": true, "score": 75.8, "grade": "BB", "agentReady": true, "rank": 47, "ranked": true, "rankOf": 1184, "categoryRank": 1, "methodology": "0.5", "run": "2026-10-01", "scores": { "ergonomics": 85, "maintenance": 75, "payments": 60, "reliability": 82, "schema": 95, "security": 55, "transparency": 74 }, "pending": [ "performance", "tasks" ], "assessment": { "confidence": "medium", "date": "2026-10-20" }, "negative": 0, "verdict": "Kolibri-1 comes with a full model card, a chat template, a tool-calling example, stated hardware needs and an open evaluation set-up, under Apache 2.0 with no acceptable use policy. It runs through Aleph Alpha's vLLM plugin or Transformers 5.20.0, about 78 GB at FP8, and no safety results are published.", "bestFor": "German and English agent work on one 80 GB-class GPU pair or a 128 GB machine, where the licence and a documented tool format matter more than raw scores.", "strengths": [ "Apache 2.0 weights with no acceptable use policy, downloadable with no account", "Safetensors only, with FP8 and BF16 builds from Aleph Alpha and about 78 GB needed at FP8 per the card", "Tool calling documented with its vLLM parser and a full example, plus reasoning effort from none to high", "262,144 tokens of native context, which the card says extends to 1,048,576", "A data card with the mix for each training stage, a June 2026 cut-off and evaluations run with an open framework" ], "weaknesses": [ "No safety evaluation results published, though the card says refusals were measured", "Runs on vLLM only through Aleph Alpha's plugin, and no Ollama, SGLang or mlx-lm support was found", "2 of 11 discussions answered by Aleph Alpha, and a report of a broken tool-call parser in the container image had no reply", "Structured output is claimed on the card without saying how to request it", "In Aleph Alpha's own table it trails several compared models on BFCL v4 multi-turn (47.5) and TerminalBench 2.1 (27.7), the vendor's numbers" ], "agentNotes": [ "Serve it with `--tool-call-parser kolibri1 --reasoning-parser kolibri1 --enable-auto-tool-choice`, or tool calls come back as text", "Install aleph-alpha-inference for vLLM 0.29, or use a Transformers release from 5.20.0", "Set `chat_template_kwargs` to `{\"enable_thinking\": false}` or `reasoning_effort: \"none\"` for short tasks, since thinking is on by default", "Keep contexts to 262,144 tokens unless you need more, as Aleph Alpha recommends, and pass the config override to go further", "Validate tool results in your own code, since the card says the trained safeguards don't replace application-level checks" ], "metrics": { "kind": "local", "measured": false }, "reviewCount": 4, "avgRating": 3.5, "history": [ { "basis": "public evidence", "confidence": "medium", "grade": "BB", "methodology": "0.5", "pending": [ "performance", "tasks" ], "run": "2026-10-01", "runLabel": "October 2026 research run", "score": 75.8 } ], "editorialScores": { "ergonomics": 85, "maintenance": 75, "payments": 60, "reliability": 82, "schema": 95, "security": 55, "transparency": 80 }, "provenanceScore": 68 }, "connect": { "install": "pip install 'aleph-alpha-inference\u003e=1'\nvllm serve Aleph-Alpha/Kolibri-1 --kv-cache-dtype fp8 \\\n --reasoning-parser kolibri1 --tool-call-parser kolibri1 --enable-auto-tool-choice", "http": "curl http://localhost:8000/v1/chat/completions \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"model\": \"Aleph-Alpha/Kolibri-1\",\n \"messages\": [{\"role\": \"user\", \"content\": \"What is the capital of France?\"}],\n \"chat_template_kwargs\": {\"enable_thinking\": false}\n }'" }, "letme": { "capability": "https://letme.dev/model.open-weights", "tool": "https://letme.dev/kolibri-1" }, "area": "models", "provenance": { "legalEntity": "Aleph Alpha GmbH", "domain": "aleph-alpha.com", "domainRegistered": "2018-08-11", "endpointOnVendorDomain": null, "terms": "", "privacy": "", "statusPage": "", "changelog": "https://huggingface.co/Aleph-Alpha/Kolibri-1/commits/main", "securityTxt": "unknown", "checked": "2026-10-20", "notes": [ "The model card names Aleph Alpha GmbH as the provider and Aleph Alpha Research GmbH as the developer. The imprint names Aleph Alpha GmbH, Heidelberg, commercial register Mannheim HRB 732804.", "Nothing is hosted, so the Apache 2.0 licence stands in for terms of service and no privacy policy is scored.", "The changelog is the repository's commit history, whose messages say what changed (Release 2026-10-02, a quantisation config fix on 13 October, transformers_version added on 19 October).", "aleph-alpha.com/.well-known/security.txt answered 403 to our reader on 20 October 2026, so it is counted as unchecked rather than absent." ], "score": 68 }, "pageJsonUrl": "https://www.anchorterminal.com/tools/kolibri-1.json", "live": { "slug": "kolibri-1", "pages": [ { "url": "https://huggingface.co/Aleph-Alpha/Kolibri-1/commits/main", "kind": "changelog", "status": 200, "checkedAt": "2026-10-21T00:49:59.261303558Z", "changedAt": "0001-01-01T00:00:00Z", "fingerprint": "037a793cc2d6" } ], "updatedAt": "2026-10-21T00:49:59.261303558Z" } }, { "slug": "qwen3-8-27b", "name": "Qwen3.8-27B", "vendor": "Alibaba (Qwen team)", "vendorUrl": "https://qwen.ai", "kind": "weights", "category": "open-weights", "summary": "Qwen3.8-27B is Alibaba's open-weight dense vision-language model of 27B parameters, with thinking on by default and a 262,144-token context, published on Hugging Face in BF16 and FP8 under Apache 2.0.", "url": "https://www.anchorterminal.com/tools/qwen3-8-27b", "markdownUrl": "https://www.anchorterminal.com/tools/qwen3-8-27b.md", "slimMarkdownUrl": "https://www.anchorterminal.com/tools/qwen3-8-27b.min.md", "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/qwen3-8-27b.json", "repo": "https://huggingface.co/Qwen/Qwen3.8-27B", "license": "Apache-2.0", "transports": [], "packages": [ { "registry": "huggingface", "name": "Qwen/Qwen3.8-27B" }, { "registry": "huggingface", "name": "Qwen/Qwen3.8-27B-FP8" }, { "registry": "ollama", "name": "qwen3.8" } ], "auth": "none", "authNotes": "The weights download from Hugging Face with no account, token or terms to accept. The repository is not gated (checked 2026-10-20).", "pricing": "free", "pricingNotes": "Free to download and use, commercial use included, under Apache 2.0. The card says a hosted version on Qwen Cloud is coming soon. Running it yourself costs the hardware, 55.6 GB for the BF16 files and 18 GB for Ollama's default build (checked 2026-10-20).", "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 2026-10-20).", "endpoints": [] }, "toolCount": null, "popularity": { "githubStars": null, "npmWeekly": null, "pypiWeekly": null, "asOf": "2026-10-20" }, "docsUrl": "https://huggingface.co/Qwen/Qwen3.8-27B", "capabilities": [ "model.open-weights" ], "tags": [ "open-weights", "model", "dense", "multimodal", "reasoning", "tool-calling", "apache-2.0", "safetensors", "fp8", "self-hosted" ], "lastRelease": "2026-08-14", "graded": true, "anchor": { "graded": true, "score": 63.7, "grade": "B", "agentReady": false, "rank": 428, "ranked": true, "rankOf": 1184, "categoryRank": 2, "methodology": "0.5", "run": "2026-10-01", "scores": { "ergonomics": 85, "maintenance": 25, "payments": 60, "reliability": 90, "schema": 63, "security": 45, "transparency": 47 }, "pending": [ "performance", "tasks" ], "assessment": { "confidence": "medium", "date": "2026-10-20" }, "negative": 0, "verdict": "Qwen3.8-27B runs nearly anywhere, in vLLM, SGLang, llama.cpp, Ollama and MLX, with an 18 GB Ollama build and Apache 2.0 terms. The card leaves out tool calling, hardware needs, training data, limitations and safety results, the files haven't changed since 14 August, and Alibaba answered none of the 40 most recent discussions.", "bestFor": "People who want a capable open model on one 24 to 32 GB GPU or Mac with the widest runtime support and a plain licence.", "strengths": [ "Model code in vLLM, SGLang, mlx-lm and Transformers 5.20.0, the architecture in llama.cpp and a local entry in Ollama's library", "Apache 2.0 weights with no acceptable use policy, downloadable with no account", "Ollama's default build is 18 GB and Alibaba's FP8 repository 30.9 GB", "262,144 tokens of native context, thinking that can be turned off per request and three reasoning effort levels", "Safetensors only, with a chat template in the repository" ], "weaknesses": [ "The card doesn't describe tool calling, so the parser to use comes from vLLM's and SGLang's recipes", "No training data, limitations, languages, knowledge cut-off or safety results on the card", "No memory needs stated by Alibaba", "None of the 40 most recent discussions had a reply from a Qwen member, including reports about the chat template", "The files and card haven't changed since 14 August 2026" ], "agentNotes": [ "Serve it with `--tool-call-parser qwen3_coder` and `--reasoning-parser qwen3`; the Hermes parser leaves its tool calls unparsed, per SGLang's cookbook", "Set `enable_thinking: false` in `chat_template_kwargs` for short tasks, since thinking is on by default", "Use the sampling the card gives for each mode, `temperature=1.0, top_p=0.95` thinking and `temperature=0.7, top_p=0.8, presence_penalty=1.5` without", "Turn on YaRN only for contexts beyond 262,144 tokens; the card says static YaRN can cost quality on short texts", "Check JSON output yourself, since structured output isn't documented for the model" ], "metrics": { "kind": "local", "measured": false }, "reviewCount": 4, "avgRating": 3, "history": [ { "basis": "public evidence", "confidence": "medium", "grade": "B", "methodology": "0.5", "pending": [ "performance", "tasks" ], "run": "2026-10-01", "runLabel": "October 2026 research run", "score": 63.7 } ], "editorialScores": { "ergonomics": 85, "maintenance": 25, "payments": 60, "reliability": 90, "schema": 63, "security": 45, "transparency": 45 }, "provenanceScore": 49 }, "connect": { "install": "ollama run qwen3.8\n# or, with vLLM's recipe:\nvllm serve Qwen/Qwen3.8-27B-FP8 --max-model-len 262144 --reasoning-parser qwen3 \\\n --enable-auto-tool-choice --tool-call-parser qwen3_coder", "http": "curl http://localhost:11434/api/chat \\\n -d '{\"model\": \"qwen3.8\", \"messages\": [{\"role\": \"user\", \"content\": \"Hello!\"}]}'" }, "letme": { "capability": "https://letme.dev/model.open-weights", "tool": "https://letme.dev/qwen3-8-27b" }, "area": "models", "provenance": { "legalEntity": "Alibaba Cloud (named in the licence's copyright line)", "domain": "qwen.ai", "domainRegistered": "2023-08-27", "endpointOnVendorDomain": null, "terms": "", "privacy": "", "statusPage": "", "changelog": "", "securityTxt": "none", "checked": "2026-10-20", "notes": [ "The LICENSE file reads Copyright 2026 Alibaba Cloud. RDAP names Alibaba Cloud (Singapore) Private Limited as the registrant of qwen.ai.", "Nothing is hosted, so the Apache 2.0 licence stands in for terms of service and no privacy policy is scored.", "No changelog. The commit history's messages are uploads and README updates that don't say what changed.", "qwen.ai answers every path, /.well-known/security.txt and /robots.txt included, with its web app, so there is no security.txt." ], "score": 49 }, "pageJsonUrl": "https://www.anchorterminal.com/tools/qwen3-8-27b.json" }, { "slug": "deepseek-v4-1-flash", "name": "DeepSeek V4.1 Flash", "vendor": "DeepSeek", "vendorUrl": "https://www.deepseek.com", "kind": "weights", "category": "open-weights", "summary": "DeepSeek V4.1 Flash is DeepSeek's open-weight multimodal mixture-of-experts model, 552B backbone parameters with 8B active in prefill and 16B in decode and a 1M-token context, published on Hugging Face under MIT. The 17 October update is graded.", "url": "https://www.anchorterminal.com/tools/deepseek-v4-1-flash", "markdownUrl": "https://www.anchorterminal.com/tools/deepseek-v4-1-flash.md", "slimMarkdownUrl": "https://www.anchorterminal.com/tools/deepseek-v4-1-flash.min.md", "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/deepseek-v4-1-flash.json", "repo": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017", "license": "MIT", "transports": [], "packages": [ { "registry": "huggingface", "name": "deepseek-ai/DeepSeek-V4.1-Flash-1017" }, { "registry": "huggingface", "name": "deepseek-ai/DeepSeek-V4.1-Flash" } ], "auth": "none", "authNotes": "The weights download from Hugging Face with no account, token or terms to accept. Neither repository is gated (checked 2026-10-20).", "pricing": "free", "pricingNotes": "Free to download and use, commercial use included, under the MIT licence. DeepSeek's own API serves the same weights as deepseek-flash, listed separately. Running them yourself costs the hardware, 510 GB of files per repository (checked 2026-10-20).", "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 2026-10-20).", "endpoints": [] }, "toolCount": null, "popularity": { "githubStars": null, "npmWeekly": null, "pypiWeekly": null, "asOf": "2026-10-20" }, "docsUrl": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017", "capabilities": [ "model.open-weights" ], "tags": [ "open-weights", "model", "moe", "multimodal", "reasoning", "tool-calling", "mit", "safetensors", "fp8", "self-hosted", "new" ], "lastRelease": "2026-10-17", "graded": true, "anchor": { "graded": true, "score": 62.8, "grade": "B", "agentReady": false, "rank": 472, "ranked": true, "rankOf": 1184, "categoryRank": 3, "methodology": "0.5", "run": "2026-10-01", "scores": { "ergonomics": 70, "maintenance": 70, "payments": 60, "reliability": 72, "schema": 69, "security": 35, "transparency": 69 }, "pending": [ "performance", "tasks" ], "assessment": { "confidence": "medium", "date": "2026-10-20" }, "negative": 0, "verdict": "DeepSeek V4.1 Flash gives a 1M-token context, trained tool calls and an effort budget from 1 to 100 under the MIT licence, with every release kept in its own repository. It needs 510 GB of files and runs in vLLM or mlx-lm, prompts come from DeepSeek's code rather than a template, and no safety results or training sources are published.", "bestFor": "Teams with a multi-GPU server who want DeepSeek's agent model under MIT with the API as a fallback.", "strengths": [ "MIT licence for the weights with no acceptable use policy, downloadable with no account", "1M tokens of context trained in, with an effort budget from 1 to 100 and a non-thinking mode", "Tool-call format documented with test cases, and DeepSeek's deepseek-recipe library parses it", "Each release kept in its own repository, and the 17 October update dated and described", "Safetensors only, with model code in vLLM and mlx-lm" ], "weaknesses": [ "510.3 GB of files, with no memory needs stated by DeepSeek", "No Jinja template in the 17 October repository, so prompts are built by running DeepSeek's encoding code", "Not in Transformers 5.20.0 or llama.cpp, and Ollama serves it from its cloud only", "No safety evaluation, training sources or knowledge cut-off published", "None of the 34 discussions listed for the September repository had a reply from DeepSeek" ], "agentNotes": [ "Build prompts with encoding.py or deepseek-recipe; the September repository's Jinja template is behind the 16 October encoding changes per an open pull request", "Parse tool calls from the DSML `calls` block, which V4.1 writes with a space before `calls`, `invoke` and `parameter`, unlike V4", "Pass reasoning effort as a number from 1 to 100, or low, high or max (50, 75, 100), in thinking mode", "Allow 256K or more output tokens at high effort, as the card recommends", "Use DeepSeek's hosted API (the deepseek-api listing) if you can't hold 510 GB of weights" ], "metrics": { "kind": "local", "measured": false }, "reviewCount": 4, "avgRating": 2.8, "history": [ { "basis": "public evidence", "confidence": "medium", "grade": "B", "methodology": "0.5", "pending": [ "performance", "tasks" ], "run": "2026-10-01", "runLabel": "October 2026 research run", "score": 62.8 } ], "editorialScores": { "ergonomics": 70, "maintenance": 70, "payments": 60, "reliability": 72, "schema": 69, "security": 35, "transparency": 65 }, "provenanceScore": 73 }, "connect": { "install": "# DeepSeek's reference demo, from the repository's inference folder (8 GPUs)\npip install -r inference/requirements.txt\npython convert.py --hf-ckpt-path \"$HF_CKPT_PATH\" --save-path \"$SAVE_PATH\" \\\n --model-parallel 8 --expert-dtype fp4 --tokenizer-path \"$HF_CKPT_PATH\"\ntorchrun --nproc-per-node 8 generate.py --ckpt-path \"$SAVE_PATH\" --config config.json --interactive" }, "letme": { "capability": "https://letme.dev/model.open-weights", "tool": "https://letme.dev/deepseek-v4-1-flash" }, "sameCompany": [ "deepseek-search-api" ], "sameModel": [ "deepseek-api" ], "area": "models", "provenance": { "legalEntity": "DeepSeek (the MIT licence names DeepSeek, and DeepSeek's platform terms name Hangzhou DeepSeek Artificial Intelligence Co., Ltd.)", "domain": "deepseek.com", "domainRegistered": "2000-08-13", "domainNote": "deepseek.com was registered in 2000, long before the company, so domain age flatters it.", "endpointOnVendorDomain": null, "terms": "", "privacy": "", "statusPage": "", "changelog": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/commits/main", "securityTxt": "none", "checked": "2026-10-20", "notes": [ "The legal entity is the one DeepSeek's platform terms name, read for the deepseek-api listing. The weights' licence names only DeepSeek.", "Nothing is hosted, so the MIT licence stands in for terms of service and no privacy policy is scored.", "The changelog is the September repository's commit history, whose messages say what changed (an encoding fix on 16 October, an indexer cache fix on 12 October, the chat template on 1 October).", "deepseek.com/.well-known/security.txt and www.deepseek.com/.well-known/security.txt answered 404 on 20 October 2026." ], "score": 73 }, "pageJsonUrl": "https://www.anchorterminal.com/tools/deepseek-v4-1-flash.json", "live": { "slug": "deepseek-v4-1-flash", "pages": [ { "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/commits/main", "kind": "changelog", "status": 200, "checkedAt": "2026-10-21T00:50:03.414291506Z", "changedAt": "2026-10-21T00:50:03.414291506Z", "fingerprint": "96f7ad6c7b35" } ], "updatedAt": "2026-10-21T00:50:03.414291506Z" } } ] }, "kind": "anchor.page", "links": { "api": "https://www.anchorterminal.com/api/v1/index.json", "html": "https://www.anchorterminal.com/categories/open-weights", "json": "https://www.anchorterminal.com/categories/open-weights.json", "llms": "https://www.anchorterminal.com/llms.txt", "markdown": "https://www.anchorterminal.com/categories/open-weights.md", "mcp": "https://www.anchorterminal.com/mcp", "slim": "https://www.anchorterminal.com/categories/open-weights.min.md" }, "markdown": "Model weights you download and run on your own hardware, under a licence that says what you may do with them. Each is graded on the files, apart from any hosted API that serves the same model. Compared on the licence, the runtimes that load it, tool calling, context, the memory it needs and what the vendor publishes about training data and evaluations.\n\n- Tools ranked: 3 · agent-ready (BB or better): 1 · accept x402: 0 · hosted endpoints: 0 · desk reviews by the panel: 12\n- JSON: https://www.anchorterminal.com/api/v1/tools.json (list) · https://www.anchorterminal.com/api/v1/rankings.json (ranked) · https://www.anchorterminal.com/api/v1/x402.json (payable) · https://www.anchorterminal.com/api/v1/capabilities.json (by capability)\n- Grades run AA, A, BB, B, C, D, E, F · methodology: https://www.anchorterminal.com/benchmark/\n\n- Head-to-head comparisons (3): https://www.anchorterminal.com/compare/open-weights/index.md\n- Capabilities in this category: model.open-weights\n- https://letme.dev/model.open-weights picks the top-graded tool in this list and says how to call it direct (https://www.anchorterminal.com/letme/index.md)\n\n## Ranking\n\n| # | Tool | Vendor | Kind | Category | Grade | Score | Confidence | x402 | Auth | Where | Reviews | Page |\n| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |\n| 47 | Kolibri-1 | Aleph Alpha | Open-weight model | Open weights | BB | 75.8 | medium | no | None | local | 3.5/5 (4) | https://www.anchorterminal.com/tools/kolibri-1.md |\n| 428 | Qwen3.8-27B | Alibaba (Qwen team) | Open-weight model | Open weights | B | 63.7 | medium | no | None | local | 3/5 (4) | https://www.anchorterminal.com/tools/qwen3-8-27b.md |\n| 472 | DeepSeek V4.1 Flash | DeepSeek | Open-weight model | Open weights | B | 62.8 | medium | no | None | local | 2.8/5 (4) | https://www.anchorterminal.com/tools/deepseek-v4-1-flash.md |\n\nScores are from public evidence against the published checklist (https://www.anchorterminal.com/benchmark/), with Performance and Task success pending. p95 latency and context cost come from our probes, which haven't run yet.\n\n## Summaries\n\n### 47. Kolibri-1, BB (75.8)\n\nKolibri-1 is Aleph Alpha's open-weight mixture-of-experts reasoning model for German and English, 78B parameters with 3.46B active per token, published on Hugging Face in FP8 and BF16 under Apache 2.0 and served with vLLM. Kolibri-1 comes with a full model card, a chat template, a tool-calling example, stated hardware needs and an open evaluation set-up, under Apache 2.0 with no acceptable use policy. It runs through Aleph Alpha's vLLM plugin or Transformers 5.20.0, about 78 GB at FP8, and no safety results are published.\n\n- Page: https://www.anchorterminal.com/tools/kolibri-1 · Markdown: https://www.anchorterminal.com/tools/kolibri-1.md · JSON: https://www.anchorterminal.com/api/v1/tools/kolibri-1.json\n- Capabilities: model.open-weights\n\n### 428. Qwen3.8-27B, B (63.7)\n\nQwen3.8-27B is Alibaba's open-weight dense vision-language model of 27B parameters, with thinking on by default and a 262,144-token context, published on Hugging Face in BF16 and FP8 under Apache 2.0. Qwen3.8-27B runs nearly anywhere, in vLLM, SGLang, llama.cpp, Ollama and MLX, with an 18 GB Ollama build and Apache 2.0 terms. The card leaves out tool calling, hardware needs, training data, limitations and safety results, the files haven't changed since 14 August, and Alibaba answered none of the 40 most recent discussions.\n\n- Page: https://www.anchorterminal.com/tools/qwen3-8-27b · Markdown: https://www.anchorterminal.com/tools/qwen3-8-27b.md · JSON: https://www.anchorterminal.com/api/v1/tools/qwen3-8-27b.json\n- Capabilities: model.open-weights\n\n### 472. DeepSeek V4.1 Flash, B (62.8)\n\nDeepSeek V4.1 Flash is DeepSeek's open-weight multimodal mixture-of-experts model, 552B backbone parameters with 8B active in prefill and 16B in decode and a 1M-token context, published on Hugging Face under MIT. The 17 October update is graded. DeepSeek V4.1 Flash gives a 1M-token context, trained tool calls and an effort budget from 1 to 100 under the MIT licence, with every release kept in its own repository. It needs 510 GB of files and runs in vLLM or mlx-lm, prompts come from DeepSeek's code rather than a template, and no safety results or training sources are published.\n\n- Page: https://www.anchorterminal.com/tools/deepseek-v4-1-flash · Markdown: https://www.anchorterminal.com/tools/deepseek-v4-1-flash.md · JSON: https://www.anchorterminal.com/api/v1/tools/deepseek-v4-1-flash.json\n- Capabilities: model.open-weights\n\n## How we test this category\n\nIn this run, public evidence against the open-weight reading of the checklist, from the model card, the files, the licence and the runtimes' own code. When our local runs start, each model that fits in 128 GB is served on one machine through llama.cpp and vLLM at the largest precision that fits. We measure tokens per second and time to first token, then run the agent task suite with the model driving. This test hasn't run yet, so Task success is pending and the grades here come from the categories assessed from public evidence.\n\n", "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": "2026-10-21", "license": "CC-BY-4.0", "method": "https://www.anchorterminal.com/benchmark/", "methodology": "0.5", "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" }, "page": { "breadcrumbs": [ { "name": "Home", "url": "https://www.anchorterminal.com/" }, { "name": "Terminal", "url": "https://www.anchorterminal.com/tools/" }, { "name": "Open-weight models", "url": "" } ], "description": "3 open-weight models ranked by the Anchor benchmark. Leader Kolibri-1 (BB). Model weights you download and run on your own hardware, under a licence that says what you may do with them. Each is graded on the files, apart from any hosted API that serves the same model. Compared on the licence, the runtimes that load it, tool calling, context, the memory it needs and what the vendor publishes about training data and evaluations.", "facts": [ "Kolibri-1 BB", "Qwen3.8-27B B", "DeepSeek V4.1 Flash B" ], "h1": "Open-weight models you download and run yourself", "image": "https://www.anchorterminal.com/assets/og/categories-open-weights.png", "path": "/categories/open-weights", "published": "", "section": "tools", "title": "Open-weight models you download and run yourself, ranked", "toc": null, "updated": "2026-10-20", "url": "https://www.anchorterminal.com/categories/open-weights" }, "tokens": { "markdown": 1450, "slim": 280 }, "version": 1 }