{ "data": { "similar": [ { "grade": "BB", "json": "https://www.anchorterminal.com/tools/openai-decisions-api.json", "name": "OpenAI Decisions API", "score": 71.5, "shared": [ "inference.decision" ], "slug": "openai-decisions-api" }, { "grade": "B", "json": "https://www.anchorterminal.com/tools/decider.json", "name": "Decider", "score": 69.5, "shared": [ "inference.decision" ], "slug": "decider" }, { "grade": "B", "json": "https://www.anchorterminal.com/tools/convai-laya.json", "name": "Laya", "score": 69.2, "shared": [ "inference.decision" ], "slug": "convai-laya" }, { "grade": "B", "json": "https://www.anchorterminal.com/tools/jaredpalmer-kev.json", "name": "Kev", "score": 67.4, "shared": [ "inference.decision" ], "slug": "jaredpalmer-kev" }, { "grade": "B", "json": "https://www.anchorterminal.com/tools/vela.json", "name": "Vela 2.0", "score": 66.5, "shared": [ "inference.decision" ], "slug": "vela" }, { "grade": "B", "json": "https://www.anchorterminal.com/tools/cloudflare-clef.json", "name": "Clef", "score": 66.1, "shared": [ "inference.decision" ], "slug": "cloudflare-clef" } ], "tool": { "slug": "gliclass", "name": "GLiClass", "vendor": "Knowledgator", "vendorUrl": "https://www.knowledgator.com", "kind": "model", "category": "decision-models", "summary": "GLiClass is an open-source Python library and family of open-weight zero-shot text classifiers from Knowledgator. It scores every candidate label in one forward pass and runs locally through a pipeline or a Ray Serve endpoint.", "url": "https://www.anchorterminal.com/tools/gliclass", "markdownUrl": "https://www.anchorterminal.com/tools/gliclass.md", "slimMarkdownUrl": "https://www.anchorterminal.com/tools/gliclass.min.md", "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/gliclass.json", "repo": "https://github.com/Knowledgator/GLiClass", "license": "Apache-2.0 (library and the model weights we checked)", "transports": [ "http" ], "packages": [ { "registry": "pypi", "name": "gliclass" } ], "auth": "none", "authNotes": "No account or key. The weights are public and ungated on Hugging Face. The bundled Ray Serve deployment has no authentication of any kind, and `python -m gliclass.serve` binds to 0.0.0.0 unless `--host` is passed, so the owner has to restrict the port.", "pricing": "free", "pricingNotes": "Free and open source, with nothing to buy for the library or the weights. Hardware is the owner's cost. Knowledgator's site links a hosted platform with plans at platform.knowledgator.com, which we couldn't reach on 8 October 2026, so whether it serves GLiClass and at what price is unchecked.", "priceSummary": "Free · OSS", "where": "local", "x402": { "level": "no", "evidence": "No x402, MPP or L402. GLiClass is software the owner runs, and its server has no payment route (checked 2026-10-08).", "endpoints": [] }, "toolCount": null, "popularity": { "githubStars": 555, "npmWeekly": null, "pypiWeekly": 13204, "asOf": "2026-10-08" }, "docsUrl": "https://docs.knowledgator.com/docs/frameworks/gliclass/", "capabilities": [ "inference.decision" ], "tags": [ "model", "open-source", "open-weights", "self-hosted", "local", "free", "python", "no-auth" ], "lastRelease": "2026-07-21", "graded": true, "anchor": { "graded": true, "score": 49.9, "grade": "D", "agentReady": false, "rank": 697, "ranked": true, "rankOf": 842, "categoryRank": 9, "methodology": "0.4", "run": "2026-10-01", "scores": { "ergonomics": 60, "maintenance": 44, "payments": 60, "reliability": 43, "schema": 49, "security": 38, "transparency": 64 }, "pending": [ "performance", "tasks" ], "breakdown": [ { "key": "reliability", "name": "Reliability", "weight": 16, "effectiveWeight": 20, "score": 43, "points": 8.6, "reason": "Scored on the local-package checklist, since GLiClass is a library and weights the owner runs. `pip install gliclass` from PyPI at 0.1.20 with Python 3.10 or newer stated. The README's source install calls a `requirements.txt` the repository doesn't have, and the base v3.0 card still names `transformers\u003e=4.48.0` while the package requires 5.0 (17 of 20). A public Tests workflow runs 151 pytest functions on Python 3.10 to 3.12 and a ruff job. Its last three runs on main, all on 24 September 2026, failed, after five passing runs from 19 May to 21 July. We didn't read which job failed (10 of 25). 15 issues and 1 pull request were open on 8 October, the oldest from 22 September 2024, among them #44 (models not loading with transformers v5, 6 July 2026), #35 (demo error, 6 February 2026) and #12 (every label scored 0.99). We didn't check reply times (12 of 25). No changelog file. Three GitHub releases carry one-line notes, and 0.1.18 moved the dependency floor to transformers 5 in a patch version (4 of 15). Version 0.1.20, with no statement of stability (0)." }, { "key": "performance", "name": "Performance", "weight": 10, "effectiveWeight": 0, "pending": true, "points": 0, "reason": "Pending. Latency is measured per call by our probes, which haven't run yet, so this run doesn't score it. Its weight is shared across the assessed categories until the first probe window closes." }, { "key": "schema", "name": "Schema \u0026 documentation", "weight": 13, "effectiveWeight": 16.25, "score": 49, "points": 7.96, "reason": "Read for a model the owner serves. No OpenAPI file. The serving page documents the seven request fields of `POST /gliclass` and the response array in a table, and the pipeline's Python signatures carry type hints (10 of 25). docs.knowledgator.com has no llms.txt (404). The README, `docs/streaming.md` and the model cards are Markdown (5 of 10). The cards and docs name the intended uses (topic, sentiment, intent, natural language inference, reranking) and print per-dataset F1 that shows the weak cases, such as 0.3376 on sst5 for base v3.0. The base card has no limitations section and says nothing on calibration (11 of 20). Labels are free strings or a dict of groups, the mode is one of two values and `threshold` is a number. The HTTP handler reads fields from the body without validation, and a list in `texts` is cut to its first item (8 of 15). Examples for every mode in the docs. The handler's 400 and 404 answers are not documented, and a request without `labels` gets a 404 (8 of 15). Versions on PyPI and version numbers in model names, with three one-line GitHub release notes and no changelog (7 of 15)." }, { "key": "ergonomics", "name": "Agent ergonomics", "weight": 13, "effectiveWeight": 16.25, "score": 60, "points": 9.75, "reason": "Read as an API an agent calls for a decision, as with the other decision models. The answer is a short list of labels and scores and no text is generated. Text and labels share one sequence, 1,024 tokens by default, with 25 labels by default (17 of 25). The caller sets a threshold, picks single-label or multi-label, and can ask for hierarchical output. The pipeline takes a batch of texts and a chunking pipeline handles long documents. The HTTP endpoint takes one text a request (15 of 20). Errors are `{\"error\": \"...\"}` with 400 or 404 and are undocumented. Truncation of long input is silent (7 of 20). Classification calls are stateless and safe to retry. The server has a request timeout and a queue capacity in its configuration. We found no `Retry-After` or retry guidance in the serve package or the docs (12 of 20). Loading takes a model, a tokeniser and a pipeline call, and a Python client ships with the server. Python is the only language, and the server defaults to CUDA and needs three flags on a CPU (9 of 15)." }, { "key": "security", "name": "Security \u0026 auth", "weight": 14, "effectiveWeight": 17.5, "score": 38, "points": 6.65, "reason": "Read as software the owner runs. No account. The Ray Serve deployment checks no credential and has no option for one, and the command line binds to 0.0.0.0 by default (6 of 30). A classifier has no write actions, and the one other route, `/adapter-cache`, only reads (15 of 20). The output is labels and scores, so no untrusted text comes back. The docs warn that the safety checkpoints can miss new or obfuscated attacks. Nothing documents how hostile text moves a score (10 of 15). The server logs start-up and label truncation only, and the docs describe no per-call record (3 of 15). No SECURITY.md, no security.txt on knowledgator.com (404) and no bounty. Weights are safetensors and PyPI releases use trusted publishing. `gliclass/config.py` loads an unknown encoder's configuration with `trust_remote_code=True` (4 of 20)." }, { "key": "payments", "name": "Payments \u0026 pricing", "weight": 10, "effectiveWeight": 12.5, "score": 60, "points": 7.5, "reason": "Free Apache-2.0 software with nothing to buy, so 20 + 20 + 20 for pricing, free use and no sign-up under the self-hosted rule. No payment protocol (0). Knowledgator's hosted platform and its plans were unreachable from our network on 8 October 2026 and are not graded here." }, { "key": "tasks", "name": "Task success", "weight": 10, "effectiveWeight": 0, "pending": true, "points": 0, "reason": "Pending. Task success needs the category task suites run through each tool, which haven't run yet, so this run doesn't score it. Its weight is shared across the assessed categories until then. A data provider's data-quality score is published on its listing now and becomes half of this category when it's scored." }, { "key": "maintenance", "name": "Maintenance \u0026 community", "weight": 7, "effectiveWeight": 8.75, "score": 44, "points": 3.85, "reason": "Read for an open-weight model with a library. 0.1.20 on 21 July 2026, 79 days before the check (20 of 30). One release in the last 90 days, and the newest GLiClass checkpoints we found on the Hub date from April 2026 (0 of 20). Three outside pull requests were merged on 24 September 2026. 15 issues were open, eleven of them more than a year old, and we didn't check reply times. A Discord server is linked (10 of 25). The Python package is current and runs on transformers 5. No other SDK (10 of 15). CI and a `uv.lock` exist, and PyPI publishing is automated on tags, but the Tests workflow is failing on main (4 of 10)." }, { "key": "transparency", "name": "Transparency \u0026 trust", "weight": 7, "effectiveWeight": 8.75, "score": 64, "points": 5.6, "note": "editorial 60, provenance 68", "reason": "Apache-2.0 for the library and the weights, with `train.py`, the training datasets named on the card and an arXiv paper (30). Self-hosted, so inputs stay on the owner's hardware, though no project page says so. The company's privacy policy of 21 September 2023 covers its website and services, not the library (18 of 30). No deprecation policy. Older model generations stay on the Hub without a notice of which are current (2 of 20). We found no telemetry code in the package and no statement either way. Weights download from the Hugging Face Hub (10 of 20)." } ], "assessment": { "date": "2026-10-08", "basis": "public evidence", "confidence": "medium", "notes": { "ergonomics": "Read as an API an agent calls for a decision, as with the other decision models. The answer is a short list of labels and scores and no text is generated. Text and labels share one sequence, 1,024 tokens by default, with 25 labels by default (17 of 25). The caller sets a threshold, picks single-label or multi-label, and can ask for hierarchical output. The pipeline takes a batch of texts and a chunking pipeline handles long documents. The HTTP endpoint takes one text a request (15 of 20). Errors are `{\"error\": \"...\"}` with 400 or 404 and are undocumented. Truncation of long input is silent (7 of 20). Classification calls are stateless and safe to retry. The server has a request timeout and a queue capacity in its configuration. We found no `Retry-After` or retry guidance in the serve package or the docs (12 of 20). Loading takes a model, a tokeniser and a pipeline call, and a Python client ships with the server. Python is the only language, and the server defaults to CUDA and needs three flags on a CPU (9 of 15).", "maintenance": "Read for an open-weight model with a library. 0.1.20 on 21 July 2026, 79 days before the check (20 of 30). One release in the last 90 days, and the newest GLiClass checkpoints we found on the Hub date from April 2026 (0 of 20). Three outside pull requests were merged on 24 September 2026. 15 issues were open, eleven of them more than a year old, and we didn't check reply times. A Discord server is linked (10 of 25). The Python package is current and runs on transformers 5. No other SDK (10 of 15). CI and a `uv.lock` exist, and PyPI publishing is automated on tags, but the Tests workflow is failing on main (4 of 10).", "payments": "Free Apache-2.0 software with nothing to buy, so 20 + 20 + 20 for pricing, free use and no sign-up under the self-hosted rule. No payment protocol (0). Knowledgator's hosted platform and its plans were unreachable from our network on 8 October 2026 and are not graded here.", "reliability": "Scored on the local-package checklist, since GLiClass is a library and weights the owner runs. `pip install gliclass` from PyPI at 0.1.20 with Python 3.10 or newer stated. The README's source install calls a `requirements.txt` the repository doesn't have, and the base v3.0 card still names `transformers\u003e=4.48.0` while the package requires 5.0 (17 of 20). A public Tests workflow runs 151 pytest functions on Python 3.10 to 3.12 and a ruff job. Its last three runs on main, all on 24 September 2026, failed, after five passing runs from 19 May to 21 July. We didn't read which job failed (10 of 25). 15 issues and 1 pull request were open on 8 October, the oldest from 22 September 2024, among them #44 (models not loading with transformers v5, 6 July 2026), #35 (demo error, 6 February 2026) and #12 (every label scored 0.99). We didn't check reply times (12 of 25). No changelog file. Three GitHub releases carry one-line notes, and 0.1.18 moved the dependency floor to transformers 5 in a patch version (4 of 15). Version 0.1.20, with no statement of stability (0).", "schema": "Read for a model the owner serves. No OpenAPI file. The serving page documents the seven request fields of `POST /gliclass` and the response array in a table, and the pipeline's Python signatures carry type hints (10 of 25). docs.knowledgator.com has no llms.txt (404). The README, `docs/streaming.md` and the model cards are Markdown (5 of 10). The cards and docs name the intended uses (topic, sentiment, intent, natural language inference, reranking) and print per-dataset F1 that shows the weak cases, such as 0.3376 on sst5 for base v3.0. The base card has no limitations section and says nothing on calibration (11 of 20). Labels are free strings or a dict of groups, the mode is one of two values and `threshold` is a number. The HTTP handler reads fields from the body without validation, and a list in `texts` is cut to its first item (8 of 15). Examples for every mode in the docs. The handler's 400 and 404 answers are not documented, and a request without `labels` gets a 404 (8 of 15). Versions on PyPI and version numbers in model names, with three one-line GitHub release notes and no changelog (7 of 15).", "security": "Read as software the owner runs. No account. The Ray Serve deployment checks no credential and has no option for one, and the command line binds to 0.0.0.0 by default (6 of 30). A classifier has no write actions, and the one other route, `/adapter-cache`, only reads (15 of 20). The output is labels and scores, so no untrusted text comes back. The docs warn that the safety checkpoints can miss new or obfuscated attacks. Nothing documents how hostile text moves a score (10 of 15). The server logs start-up and label truncation only, and the docs describe no per-call record (3 of 15). No SECURITY.md, no security.txt on knowledgator.com (404) and no bounty. Weights are safetensors and PyPI releases use trusted publishing. `gliclass/config.py` loads an unknown encoder's configuration with `trust_remote_code=True` (4 of 20).", "transparency": "Apache-2.0 for the library and the weights, with `train.py`, the training datasets named on the card and an arXiv paper (30). Self-hosted, so inputs stay on the owner's hardware, though no project page says so. The company's privacy policy of 21 September 2023 covers its website and services, not the library (18 of 30). No deprecation policy. Older model generations stay on the Hub without a notice of which are current (2 of 20). We found no telemetry code in the package and no statement either way. Weights download from the Hugging Face Hub (10 of 20)." }, "sources": [ { "what": "repository at commit 68132de (README, pyproject.toml, serve package, tests, workflows)", "url": "https://github.com/Knowledgator/GLiClass", "seen": "2026-10-08" }, { "what": "PyPI metadata and release dates", "url": "https://pypi.org/pypi/gliclass/json", "seen": "2026-10-08" }, { "what": "base v3.0 model card, config and file list", "url": "https://huggingface.co/knowledgator/gliclass-base-v3.0", "seen": "2026-10-08" }, { "what": "list of Knowledgator's GLiClass repositories on the Hub", "url": "https://huggingface.co/api/models?author=knowledgator\u0026search=gliclass\u0026limit=100", "seen": "2026-10-08" }, { "what": "serving documentation", "url": "https://docs.knowledgator.com/docs/frameworks/gliclass/serving/", "seen": "2026-10-08" }, { "what": "usage, installation and pretrained-models documentation", "url": "https://docs.knowledgator.com/docs/frameworks/gliclass/usage/", "seen": "2026-10-08" }, { "what": "Tests workflow runs on main", "url": "https://github.com/Knowledgator/GLiClass/actions/workflows/tests.yml?query=branch%3Amain", "seen": "2026-10-08" }, { "what": "open issues", "url": "https://github.com/Knowledgator/GLiClass/issues", "seen": "2026-10-08" }, { "what": "GitHub releases", "url": "https://github.com/Knowledgator/GLiClass/releases", "seen": "2026-10-08" }, { "what": "terms of use (21 September 2023)", "url": "https://policies.knowledgator.com/terms-of-use.pdf", "seen": "2026-10-08" }, { "what": "privacy policy (21 September 2023)", "url": "https://policies.knowledgator.com/privacy-policy.pdf", "seen": "2026-10-08" }, { "what": "security.txt (404)", "url": "https://www.knowledgator.com/.well-known/security.txt", "seen": "2026-10-08" }, { "what": "RDAP record for knowledgator.com", "url": "https://rdap.verisign.com/com/v1/domain/knowledgator.com", "seen": "2026-10-08" }, { "what": "weekly downloads", "url": "https://pypistats.org/api/packages/gliclass/recent", "seen": "2026-10-08" } ], "openQuestions": [ "unchecked: platform.knowledgator.com (the hosted platform and its plans page). Our network's proxy answered 502 for that host, so whether GLiClass is sold as a hosted API, and at what price, is not established.", "unchecked: which job (pytest or ruff) failed in the three Tests runs of 24 September 2026. The run list was read, the logs were not.", "unchecked: reply times on issues and pull requests. The GitHub API refused us for its rate limit, and we read the issue list from the web page only.", "The lead named ONNX exports. We found no ONNX file in `knowledgator/gliclass-base-v3.0`, none among Knowledgator's 31 GLiClass repositories by name, and no export command in the library. Community ONNX builds under other accounts were not searched.", "The lead's statement that OpenClaw's docs name GLiClass as a local option was not checked.", "Licences were read for the library and `gliclass-base-v3.0` only, not for each of the 31 checkpoints.", "The fit with decision models is loose. GLiClass returns softmax or sigmoid label scores and publishes F1 only, with no expected calibration error or Brier score.", "`provenance.terms` and `provenance.privacy` are left out. Knowledgator's terms of use and privacy policy cover its website and hosted services, and no document other than the Apache-2.0 licence governs the library." ] }, "negative": 0, "verdict": "An Apache-2.0 classifier that scores a whole label set in one encoder pass on the owner's hardware, with single-label, multi-label, hierarchical and few-shot modes. It returns label scores with no calibration claim, the bundled server has no authentication, and the last three test runs on the main branch, on 24 September 2026, failed.", "bestFor": "Topic, intent and sentiment routing over a known label set on the owner's own CPU or GPU, where many labels must be scored at once.", "strengths": [ "Apache-2.0 code and weights, with `train.py`, the training datasets named on the model cards and an arXiv paper (2508.07662)", "One forward pass scores every label. The base v3.0 card reports 51.6 examples a second averaged over 1 to 128 labels on an A6000 (vendor figures)", "Single-label (softmax) and multi-label (sigmoid) modes, hierarchical label sets, task prompts and few-shot examples in one pipeline call", "A Ray Serve deployment with dynamic batching ships in the `gliclass[serve]` extra, with a documented request table for `POST /gliclass`", "31 public, ungated GLiClass checkpoints on Hugging Face, from 32.7M to 439M parameters, as safetensors" ], "weaknesses": [ "No calibration evidence. The cards report F1 only, and the docs tell users to calibrate thresholds on their own traffic", "The bundled server has no authentication, and `python -m gliclass.serve` binds to 0.0.0.0 by default", "The last three runs of the Tests workflow on main, all on 24 September 2026, failed", "Still 0.1.x with no changelog file. 0.1.18 raised the `transformers` floor to 5.0, and issue #44 on v5 loading has been open since 6 July 2026", "No OpenAPI file, no llms.txt, no SECURITY.md and no documented error responses" ], "agentNotes": [ "Pass `--host 127.0.0.1` to `python -m gliclass.serve`, or put the port behind your own gateway. The server checks no credential", "On a machine without a GPU add `--device cpu --dtype float32 --num-gpus-per-replica 0`. The default configuration expects CUDA", "Send one text a request to `POST /gliclass`. An array in `texts` is cut to its first item without an error", "Set `multi_label` to false for one label from a set. The default scores each label independently, so scores do not sum to 1", "Keep text plus labels under the pipeline's 1,024-token `max_length`, or use `ZeroShotClassificationWithChunkingPipeline`. Longer input is truncated silently" ], "metrics": { "kind": "local", "measured": false }, "reviewCount": 0, "avgRating": 0, "history": [ { "basis": "public evidence", "confidence": "medium", "grade": "D", "methodology": "0.4", "pending": [ "performance", "tasks" ], "run": "2026-10-01", "runLabel": "October 2026 research run", "score": 49.9 } ], "editorialScores": { "ergonomics": 60, "maintenance": 44, "payments": 60, "reliability": 43, "schema": 49, "security": 38, "transparency": 60 }, "provenanceScore": 68 }, "connect": { "install": "pip install gliclass", "http": "pip install \"gliclass[serve]\"\npython -m gliclass.serve --model knowledgator/gliclass-edge-v3.0 --port 8000\ncurl -X POST http://localhost:8000/gliclass \\\n -H \"Content-Type: application/json\" \\\n -d '{\"text\": \"This is a great product.\", \"labels\": [\"positive\", \"negative\", \"neutral\"], \"threshold\": 0.3, \"multi_label\": true}'" }, "letme": { "capability": "https://letme.dev/inference.decision", "tool": "https://letme.dev/gliclass" }, "notable": [ "The library is `gliclass` 0.1.20 on PyPI, released 21 July 2026, Python 3.10 or newer, depending on `torch\u003e=2.0.0` and `transformers\u003e=5.0.0` (https://pypi.org/project/gliclass/)", "`knowledgator/gliclass-base-v3.0` is a 187M-parameter uni-encoder on `microsoft/deberta-v3-base`, Apache-2.0, ungated, with 19,231 downloads in the last month per the Hub (https://huggingface.co/knowledgator/gliclass-base-v3.0)", "The card reports zero-shot F1 on 14 datasets, an average of 0.6764 for base v3.0, from 0.3376 on sst5 to 0.9474 on snips. It reports no calibration measure (https://huggingface.co/knowledgator/gliclass-base-v3.0)", "The Ray Serve deployment answers `POST /gliclass` with a JSON array of `{label, score}` and processes one text a request (https://docs.knowledgator.com/docs/frameworks/gliclass/serving/)", "The Tests workflow failed on its last three runs on main, all on 24 September 2026, after five passing runs from 19 May to 21 July (https://github.com/Knowledgator/GLiClass/actions/workflows/tests.yml)", "Issue #47 of 20 September 2026 asks for a TypeSafe-compatible endpoint and was open on 8 October (https://github.com/Knowledgator/GLiClass/issues)", "No ONNX file is in the base v3.0 repository and the library has no export command. Issue #8 asking for an ONNX build has been open since 22 September 2024" ], "area": "models", "details": [ { "label": "Library", "value": "`gliclass` 0.1.20 on PyPI (21 July 2026), Python 3.10 or newer, `torch\u003e=2.0.0`, `transformers\u003e=5.0.0`, `scikit-learn`, `numpy\u003e=2.0.0`. Extras `serve` (Ray Serve) and `streaming`" }, { "label": "Models", "value": "31 `knowledgator/gliclass-*` repositories on Hugging Face. v3.0 family of July 2025 (edge 32.7M, modern-base 151M, base 187M, modern-large 399M, large 439M), Instruct v1.0 of February 2026 and Multilang of April 2026" }, { "label": "Licence", "value": "Apache-2.0 for the library and for `gliclass-base-v3.0`, per the LICENSE file and the model card" }, { "label": "Question types", "value": "Single-label (softmax over the labels) or multi-label (sigmoid on each label), hierarchical label sets as a dict, a task prompt, few-shot examples and retrieval-augmented examples on models trained for them. No ordered rubric type" }, { "label": "Limits", "value": "Pipeline defaults of `max_classes=25` and `max_length=1024` tokens for text and labels together, with truncation. `ZeroShotClassificationWithChunkingPipeline` splits longer documents" }, { "label": "Serving", "value": "`python -m gliclass.serve` starts Ray Serve on port 8000 at `/gliclass`, one text a request, with dynamic batching. Defaults to CUDA. `GLiClassClient` and an in-process `GLiClassFactory` in Python" }, { "label": "Errors", "value": "The handler answers `{\"error\": \"...\"}` with 400 for a ValueError and 404 for a KeyError, which includes a request without `labels`. Not documented" }, { "label": "Vendor benchmarks", "value": "Base v3.0 card, zero-shot F1 averaged over 14 datasets, 0.6764, and 51.61 examples a second on an A6000 at batch size 1. Vendor figures, not ours" }, { "label": "Tests", "value": "151 pytest functions in 7 files, run on Python 3.10, 3.11 and 3.12 with a ruff job. The last three runs on main failed on 24 September 2026" }, { "label": "Releases", "value": "20 versions on PyPI since 0.1.0 on 2 June 2024. Three GitHub releases with one-line notes (0.1.18, 0.1.19, 0.1.20). No changelog file" } ], "provenance": { "legalEntity": "Knowledgator Engineering Ltd.", "domain": "knowledgator.com", "domainRegistered": "2021-06-25", "endpointOnVendorDomain": null, "terms": "", "privacy": "", "statusPage": "", "changelog": "https://github.com/Knowledgator/GLiClass/releases", "securityTxt": "none", "checked": "2026-10-08", "notes": [ "The terms of use and privacy policy linked from knowledgator.com name Knowledgator Engineering Ltd., London, United Kingdom. Both are dated 21 September 2023.", "Those two documents govern the website and Knowledgator's hosted services, not the library, so `terms` and `privacy` are left out. The Apache-2.0 licence governs what an agent would run.", "Software the owner runs, so there is no hosted endpoint and no status page for it.", "www.knowledgator.com/.well-known/security.txt returns 404, and the repository has no SECURITY.md.", "RDAP for knowledgator.com gives a registration date of 2021-06-25.", "The code is on github.com under the Knowledgator organisation and the weights on huggingface.co under knowledgator." ], "score": 68, "checks": [ { "check": "Legal entity named", "value": "Knowledgator Engineering Ltd.", "points": 20, "max": 20, "state": "ok" }, { "check": "Domain age", "value": "knowledgator.com, registered 2021-06-25 (5 years)", "points": 11, "max": 15, "state": "part" }, { "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 Apache-2.0 (library and the model weights we checked) 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" } ] }, "pageJsonUrl": "https://www.anchorterminal.com/tools/gliclass.json" }, "verify": { "accepts": "a page on knowledgator.com or one of its subdomains, or the README of github.com/Knowledgator/GLiClass", "badgeUrl": "https://www.anchorterminal.com/badges/gliclass.svg", "body": { "slug": "gliclass", "url": "the page with the badge or the link" }, "docs": "https://www.anchorterminal.com/builders/#verify", "effect": "none, it never changes a grade, rank or review", "endpoint": "https://www.anchorterminal.com/api/v1/verify", "listingUrl": "https://www.anchorterminal.com/tools/gliclass", "mcpTool": "verify_listing", "recheck": "weekly; two failed checks in a row and it lapses, a later pass restores it", "snippets": { "html": "\u003ca href=\"https://www.anchorterminal.com/tools/gliclass\"\u003e\u003cimg src=\"https://www.anchorterminal.com/badges/gliclass.svg\" alt=\"GLiClass on Anchor Terminal\" height=\"20\"\u003e\u003c/a\u003e", "markdown": "[![GLiClass on Anchor Terminal](https://www.anchorterminal.com/badges/gliclass.svg)](https://www.anchorterminal.com/tools/gliclass)", "link": "\u003ca href=\"https://www.anchorterminal.com/tools/gliclass\"\u003eGLiClass on Anchor Terminal\u003c/a\u003e" } } }, "kind": "anchor.page", "links": { "api": "https://www.anchorterminal.com/api/v1/index.json", "html": "https://www.anchorterminal.com/tools/gliclass", "json": "https://www.anchorterminal.com/tools/gliclass.json", "llms": "https://www.anchorterminal.com/llms.txt", "markdown": "https://www.anchorterminal.com/tools/gliclass.md", "slim": "https://www.anchorterminal.com/tools/gliclass.min.md" }, "markdown": "## Overview\n\n**Grade D · 49.9/100 · rank #697 of 842 · #9 in Decision models · not agent-ready · confidence medium**\n\n\n## Assessment\n\nAn Apache-2.0 classifier that scores a whole label set in one encoder pass on the owner's hardware, with single-label, multi-label, hierarchical and few-shot modes. It returns label scores with no calibration claim, the bundled server has no authentication, and the last three test runs on the main branch, on 24 September 2026, failed.\n\n## Facts\n\n| Field | Value |\n| --- | --- |\n| Vendor | Knowledgator (https://www.knowledgator.com) |\n| Kind | Model API |\n| Category | Decision models (https://www.anchorterminal.com/categories/decision-models) |\n| Transport | HTTP |\n| Auth | None · No account or key. The weights are public and ungated on Hugging Face. The bundled Ray Serve deployment has no authentication of any kind, and `python -m gliclass.serve` binds to 0.0.0.0 unless `--host` is passed, so the owner has to restrict the port. |\n| Pricing | Free (Free · OSS) · Free and open source, with nothing to buy for the library or the weights. Hardware is the owner's cost. Knowledgator's site links a hosted platform with plans at platform.knowledgator.com, which we couldn't reach on 8 October 2026, so whether it serves GLiClass and at what price is unchecked. |\n| x402 | No · No x402, MPP or L402. GLiClass is software the owner runs, and its server has no payment route (checked 2026-10-08). |\n| Licence | Apache-2.0 (library and the model weights we checked) |\n| Packages | pypi: `gliclass` |\n| Source | https://github.com/Knowledgator/GLiClass |\n| Docs | https://docs.knowledgator.com/docs/frameworks/gliclass/ |\n| llms.txt | not found |\n| Last release | 2026-07-21 |\n| GitHub stars | 555 (as of 2026-10-08) |\n| PyPI downloads / week | 13,204 |\n| Library | `gliclass` 0.1.20 on PyPI (21 July 2026), Python 3.10 or newer, `torch\u003e=2.0.0`, `transformers\u003e=5.0.0`, `scikit-learn`, `numpy\u003e=2.0.0`. Extras `serve` (Ray Serve) and `streaming` |\n| Models | 31 `knowledgator/gliclass-*` repositories on Hugging Face. v3.0 family of July 2025 (edge 32.7M, modern-base 151M, base 187M, modern-large 399M, large 439M), Instruct v1.0 of February 2026 and Multilang of April 2026 |\n| Licence | Apache-2.0 for the library and for `gliclass-base-v3.0`, per the LICENSE file and the model card |\n| Question types | Single-label (softmax over the labels) or multi-label (sigmoid on each label), hierarchical label sets as a dict, a task prompt, few-shot examples and retrieval-augmented examples on models trained for them. No ordered rubric type |\n| Limits | Pipeline defaults of `max_classes=25` and `max_length=1024` tokens for text and labels together, with truncation. `ZeroShotClassificationWithChunkingPipeline` splits longer documents |\n| Serving | `python -m gliclass.serve` starts Ray Serve on port 8000 at `/gliclass`, one text a request, with dynamic batching. Defaults to CUDA. `GLiClassClient` and an in-process `GLiClassFactory` in Python |\n| Errors | The handler answers `{\"error\": \"...\"}` with 400 for a ValueError and 404 for a KeyError, which includes a request without `labels`. Not documented |\n| Vendor benchmarks | Base v3.0 card, zero-shot F1 averaged over 14 datasets, 0.6764, and 51.61 examples a second on an A6000 at batch size 1. Vendor figures, not ours |\n| Tests | 151 pytest functions in 7 files, run on Python 3.10, 3.11 and 3.12 with a ruff job. The last three runs on main failed on 24 September 2026 |\n| Releases | 20 versions on PyPI since 0.1.0 on 2 June 2024. Three GitHub releases with one-line notes (0.1.18, 0.1.19, 0.1.20). No changelog file |\n| Capabilities | inference.decision |\n| Tags | model, open-source, open-weights, self-hosted, local, free, python, no-auth |\n| JSON | https://www.anchorterminal.com/api/v1/tools/gliclass.json |\n\n## Score breakdown (methodology v0.4, October 2026 research run)\n\nAssessed 2026-10-08 from public evidence against the published checklist (https://www.anchorterminal.com/benchmark/#checklist). Confidence: medium. Performance and Task success pending (no score, not in the total); the total is Σ(score × weight) ÷ 80 over the 7 assessed categories. \"This run\" is each category's share of the 100 points.\n\n| Category | Weight | This run | Score (0–100) | Points |\n| --- | --- | --- | --- | --- |\n| Reliability | 16% | 20 | 43 | 8.6 |\n| Performance | 10% | pending | pending | n/a |\n| Schema \u0026 documentation | 13% | 16.2 | 49 | 8.0 |\n| Agent ergonomics | 13% | 16.2 | 60 | 9.8 |\n| Security \u0026 auth | 14% | 17.5 | 38 | 6.7 |\n| Payments \u0026 pricing | 10% | 12.5 | 60 | 7.5 |\n| Task success | 10% | pending | pending | n/a |\n| Maintenance \u0026 community | 7% | 8.8 | 44 | 3.9 |\n| Transparency \u0026 trust (editorial 60, provenance 68) | 7% | 8.8 | 64 | 5.6 |\n| Negative events | up to −15 | up to −15 | none recorded | 0 |\n| **Total** | | | | **49.9 → D** |\n\n### Why each score\n\n- Reliability 43: Scored on the local-package checklist, since GLiClass is a library and weights the owner runs. `pip install gliclass` from PyPI at 0.1.20 with Python 3.10 or newer stated. The README's source install calls a `requirements.txt` the repository doesn't have, and the base v3.0 card still names `transformers\u003e=4.48.0` while the package requires 5.0 (17 of 20). A public Tests workflow runs 151 pytest functions on Python 3.10 to 3.12 and a ruff job. Its last three runs on main, all on 24 September 2026, failed, after five passing runs from 19 May to 21 July. We didn't read which job failed (10 of 25). 15 issues and 1 pull request were open on 8 October, the oldest from 22 September 2024, among them #44 (models not loading with transformers v5, 6 July 2026), #35 (demo error, 6 February 2026) and #12 (every label scored 0.99). We didn't check reply times (12 of 25). No changelog file. Three GitHub releases carry one-line notes, and 0.1.18 moved the dependency floor to transformers 5 in a patch version (4 of 15). Version 0.1.20, with no statement of stability (0).\n- Performance: Pending. Latency is measured per call by our probes, which haven't run yet, so this run doesn't score it. Its weight is shared across the assessed categories until the first probe window closes.\n- Schema \u0026 documentation 49: Read for a model the owner serves. No OpenAPI file. The serving page documents the seven request fields of `POST /gliclass` and the response array in a table, and the pipeline's Python signatures carry type hints (10 of 25). docs.knowledgator.com has no llms.txt (404). The README, `docs/streaming.md` and the model cards are Markdown (5 of 10). The cards and docs name the intended uses (topic, sentiment, intent, natural language inference, reranking) and print per-dataset F1 that shows the weak cases, such as 0.3376 on sst5 for base v3.0. The base card has no limitations section and says nothing on calibration (11 of 20). Labels are free strings or a dict of groups, the mode is one of two values and `threshold` is a number. The HTTP handler reads fields from the body without validation, and a list in `texts` is cut to its first item (8 of 15). Examples for every mode in the docs. The handler's 400 and 404 answers are not documented, and a request without `labels` gets a 404 (8 of 15). Versions on PyPI and version numbers in model names, with three one-line GitHub release notes and no changelog (7 of 15).\n- Agent ergonomics 60: Read as an API an agent calls for a decision, as with the other decision models. The answer is a short list of labels and scores and no text is generated. Text and labels share one sequence, 1,024 tokens by default, with 25 labels by default (17 of 25). The caller sets a threshold, picks single-label or multi-label, and can ask for hierarchical output. The pipeline takes a batch of texts and a chunking pipeline handles long documents. The HTTP endpoint takes one text a request (15 of 20). Errors are `{\"error\": \"...\"}` with 400 or 404 and are undocumented. Truncation of long input is silent (7 of 20). Classification calls are stateless and safe to retry. The server has a request timeout and a queue capacity in its configuration. We found no `Retry-After` or retry guidance in the serve package or the docs (12 of 20). Loading takes a model, a tokeniser and a pipeline call, and a Python client ships with the server. Python is the only language, and the server defaults to CUDA and needs three flags on a CPU (9 of 15).\n- Security \u0026 auth 38: Read as software the owner runs. No account. The Ray Serve deployment checks no credential and has no option for one, and the command line binds to 0.0.0.0 by default (6 of 30). A classifier has no write actions, and the one other route, `/adapter-cache`, only reads (15 of 20). The output is labels and scores, so no untrusted text comes back. The docs warn that the safety checkpoints can miss new or obfuscated attacks. Nothing documents how hostile text moves a score (10 of 15). The server logs start-up and label truncation only, and the docs describe no per-call record (3 of 15). No SECURITY.md, no security.txt on knowledgator.com (404) and no bounty. Weights are safetensors and PyPI releases use trusted publishing. `gliclass/config.py` loads an unknown encoder's configuration with `trust_remote_code=True` (4 of 20).\n- Payments \u0026 pricing 60: Free Apache-2.0 software with nothing to buy, so 20 + 20 + 20 for pricing, free use and no sign-up under the self-hosted rule. No payment protocol (0). Knowledgator's hosted platform and its plans were unreachable from our network on 8 October 2026 and are not graded here.\n- Task success: Pending. Task success needs the category task suites run through each tool, which haven't run yet, so this run doesn't score it. Its weight is shared across the assessed categories until then. A data provider's data-quality score is published on its listing now and becomes half of this category when it's scored.\n- Maintenance \u0026 community 44: Read for an open-weight model with a library. 0.1.20 on 21 July 2026, 79 days before the check (20 of 30). One release in the last 90 days, and the newest GLiClass checkpoints we found on the Hub date from April 2026 (0 of 20). Three outside pull requests were merged on 24 September 2026. 15 issues were open, eleven of them more than a year old, and we didn't check reply times. A Discord server is linked (10 of 25). The Python package is current and runs on transformers 5. No other SDK (10 of 15). CI and a `uv.lock` exist, and PyPI publishing is automated on tags, but the Tests workflow is failing on main (4 of 10).\n- Transparency \u0026 trust 64: Apache-2.0 for the library and the weights, with `train.py`, the training datasets named on the card and an arXiv paper (30). Self-hosted, so inputs stay on the owner's hardware, though no project page says so. The company's privacy policy of 21 September 2023 covers its website and services, not the library (18 of 30). No deprecation policy. Older model generations stay on the Hub without a notice of which are current (2 of 20). We found no telemetry code in the package and no statement either way. Weights download from the Hugging Face Hub (10 of 20).\n\nFix list for a coding agent, everything this grade says the listing lacks, the biggest gain first (18 items): https://www.anchorterminal.com/fixes/gliclass.md (JSON https://www.anchorterminal.com/fixes/gliclass.json)\n\n### What we couldn't check\n\n- unchecked: platform.knowledgator.com (the hosted platform and its plans page). Our network's proxy answered 502 for that host, so whether GLiClass is sold as a hosted API, and at what price, is not established.\n- unchecked: which job (pytest or ruff) failed in the three Tests runs of 24 September 2026. The run list was read, the logs were not.\n- unchecked: reply times on issues and pull requests. The GitHub API refused us for its rate limit, and we read the issue list from the web page only.\n- The lead named ONNX exports. We found no ONNX file in `knowledgator/gliclass-base-v3.0`, none among Knowledgator's 31 GLiClass repositories by name, and no export command in the library. Community ONNX builds under other accounts were not searched.\n- The lead's statement that OpenClaw's docs name GLiClass as a local option was not checked.\n- Licences were read for the library and `gliclass-base-v3.0` only, not for each of the 31 checkpoints.\n- The fit with decision models is loose. GLiClass returns softmax or sigmoid label scores and publishes F1 only, with no expected calibration error or Brier score.\n- `provenance.terms` and `provenance.privacy` are left out. Knowledgator's terms of use and privacy policy cover its website and hosted services, and no document other than the Apache-2.0 licence governs the library.\n\n### Sources\n\n- repository at commit 68132de (README, pyproject.toml, serve package, tests, workflows): \u003chttps://github.com/Knowledgator/GLiClass\u003e (seen 2026-10-08)\n- PyPI metadata and release dates: \u003chttps://pypi.org/pypi/gliclass/json\u003e (seen 2026-10-08)\n- base v3.0 model card, config and file list: \u003chttps://huggingface.co/knowledgator/gliclass-base-v3.0\u003e (seen 2026-10-08)\n- list of Knowledgator's GLiClass repositories on the Hub: \u003chttps://huggingface.co/api/models?author=knowledgator\u0026search=gliclass\u0026limit=100\u003e (seen 2026-10-08)\n- serving documentation: \u003chttps://docs.knowledgator.com/docs/frameworks/gliclass/serving/\u003e (seen 2026-10-08)\n- usage, installation and pretrained-models documentation: \u003chttps://docs.knowledgator.com/docs/frameworks/gliclass/usage/\u003e (seen 2026-10-08)\n- Tests workflow runs on main: \u003chttps://github.com/Knowledgator/GLiClass/actions/workflows/tests.yml?query=branch%3Amain\u003e (seen 2026-10-08)\n- open issues: \u003chttps://github.com/Knowledgator/GLiClass/issues\u003e (seen 2026-10-08)\n- GitHub releases: \u003chttps://github.com/Knowledgator/GLiClass/releases\u003e (seen 2026-10-08)\n- terms of use (21 September 2023): \u003chttps://policies.knowledgator.com/terms-of-use.pdf\u003e (seen 2026-10-08)\n- privacy policy (21 September 2023): \u003chttps://policies.knowledgator.com/privacy-policy.pdf\u003e (seen 2026-10-08)\n- security.txt (404): \u003chttps://www.knowledgator.com/.well-known/security.txt\u003e (seen 2026-10-08)\n- RDAP record for knowledgator.com: \u003chttps://rdap.verisign.com/com/v1/domain/knowledgator.com\u003e (seen 2026-10-08)\n- weekly downloads: \u003chttps://pypistats.org/api/packages/gliclass/recent\u003e (seen 2026-10-08)\n\n## Who's behind it (provenance 68/100, checked 2026-10-08)\n\n| Check | Finding | Points |\n| --- | --- | --- |\n| Legal entity named | Knowledgator Engineering Ltd. | 20/20 |\n| Domain age | knowledgator.com, registered 2021-06-25 (5 years) | 11/15 |\n| Endpoint on the vendor's domain | no hosted endpoint | n/a |\n| Terms of service | nothing hosted, so the Apache-2.0 (library and the model weights we checked) licence stands in | 10/10 |\n| Privacy policy | nothing hosted, not scored | n/a |\n| Status page | not found | 0/10 |\n| Changelog | published | 10/10 |\n| security.txt | not found | 0/10 |\n\nThe terms of use and privacy policy linked from knowledgator.com name Knowledgator Engineering Ltd., London, United Kingdom. Both are dated 21 September 2023.\n\nThose two documents govern the website and Knowledgator's hosted services, not the library, so `terms` and `privacy` are left out. The Apache-2.0 licence governs what an agent would run.\n\nSoftware the owner runs, so there is no hosted endpoint and no status page for it.\n\nwww.knowledgator.com/.well-known/security.txt returns 404, and the repository has no SECURITY.md.\n\nRDAP for knowledgator.com gives a registration date of 2021-06-25.\n\nThe code is on github.com under the Knowledgator organisation and the weights on huggingface.co under knowledgator.\n\n### Terms and privacy, as read\n\nA reading by a fixed set of rules, each answered with the vendor's own sentence. Not legal advice.\n\n**Terms of service**. Nothing is hosted by the vendor, so there are no terms of service to read. The Apache-2.0 (library and the model weights we checked) licence stands in and the check scores in full.\n\n\n**Privacy policy**. Nothing is hosted by the vendor, so there is no privacy policy to read and the check isn't scored.\n\n\n## Probe metrics\n\nNot measured yet. Our benchmark probes haven't run, so there's no availability, latency or error rate from a run and Performance is pending. Live uptime, where we poll the endpoint, is under Live and doesn't change the score.\n\n## Strengths\n\n- Apache-2.0 code and weights, with `train.py`, the training datasets named on the model cards and an arXiv paper (2508.07662)\n- One forward pass scores every label. The base v3.0 card reports 51.6 examples a second averaged over 1 to 128 labels on an A6000 (vendor figures)\n- Single-label (softmax) and multi-label (sigmoid) modes, hierarchical label sets, task prompts and few-shot examples in one pipeline call\n- A Ray Serve deployment with dynamic batching ships in the `gliclass[serve]` extra, with a documented request table for `POST /gliclass`\n- 31 public, ungated GLiClass checkpoints on Hugging Face, from 32.7M to 439M parameters, as safetensors\n\n## Weaknesses\n\n- No calibration evidence. The cards report F1 only, and the docs tell users to calibrate thresholds on their own traffic\n- The bundled server has no authentication, and `python -m gliclass.serve` binds to 0.0.0.0 by default\n- The last three runs of the Tests workflow on main, all on 24 September 2026, failed\n- Still 0.1.x with no changelog file. 0.1.18 raised the `transformers` floor to 5.0, and issue #44 on v5 loading has been open since 6 July 2026\n- No OpenAPI file, no llms.txt, no SECURITY.md and no documented error responses\n\n## Before you call it (notes for agents)\n\n1. Pass `--host 127.0.0.1` to `python -m gliclass.serve`, or put the port behind your own gateway. The server checks no credential\n2. On a machine without a GPU add `--device cpu --dtype float32 --num-gpus-per-replica 0`. The default configuration expects CUDA\n3. Send one text a request to `POST /gliclass`. An array in `texts` is cut to its first item without an error\n4. Set `multi_label` to false for one label from a set. The default scores each label independently, so scores do not sum to 1\n5. Keep text plus labels under the pipeline's 1,024-token `max_length`, or use `ZeroShotClassificationWithChunkingPipeline`. Longer input is truncated silently\n\n## Connect\n\nInstall:\n\n```bash\npip install gliclass\n```\n\nFirst request:\n\n```bash\npip install \"gliclass[serve]\"\npython -m gliclass.serve --model knowledgator/gliclass-edge-v3.0 --port 8000\ncurl -X POST http://localhost:8000/gliclass \\\n -H \"Content-Type: application/json\" \\\n -d '{\"text\": \"This is a great product.\", \"labels\": [\"positive\", \"negative\", \"neutral\"], \"threshold\": 0.3, \"multi_label\": true}'\n```\n\n## Similar tools\n\nRanked by shared capabilities, then score. Same-category tools with no shared capability key are listed last.\n\n| Tool | Grade | Score | Rank | Shared capabilities | x402 | Markdown |\n| --- | --- | --- | --- | --- | --- | --- |\n| OpenAI Decisions API | BB | 71.5 | 119 | inference.decision | no | https://www.anchorterminal.com/tools/openai-decisions-api.md |\n| Decider | B | 69.5 | 172 | inference.decision | no | https://www.anchorterminal.com/tools/decider.md |\n| Laya | B | 69.2 | 183 | inference.decision | no | https://www.anchorterminal.com/tools/convai-laya.md |\n| Kev | B | 67.4 | 235 | inference.decision | no | https://www.anchorterminal.com/tools/jaredpalmer-kev.md |\n| Vela 2.0 | B | 66.5 | 268 | inference.decision | no | https://www.anchorterminal.com/tools/vela.md |\n| Clef | B | 66.1 | 276 | inference.decision | no | https://www.anchorterminal.com/tools/cloudflare-clef.md |\n\n## Panel reviews (0)\n\nReviewed by the Anchor panel (https://www.anchorterminal.com/reviewers/index.md): .\n\nDesk reviews, written from public documentation, pricing, terms, source and status history on 1 October 2026. No calls made. For a desk review, the outcome says whether the reviewer's questions could be answered from public material: success, partial or failure. How reviews work: https://www.anchorterminal.com/reviews/how-it-works.md\n\n## Notable\n\n- The library is `gliclass` 0.1.20 on PyPI, released 21 July 2026, Python 3.10 or newer, depending on `torch\u003e=2.0.0` and `transformers\u003e=5.0.0` (source: \u003chttps://pypi.org/project/gliclass/\u003e)\n- `knowledgator/gliclass-base-v3.0` is a 187M-parameter uni-encoder on `microsoft/deberta-v3-base`, Apache-2.0, ungated, with 19,231 downloads in the last month per the Hub (source: \u003chttps://huggingface.co/knowledgator/gliclass-base-v3.0\u003e)\n- The card reports zero-shot F1 on 14 datasets, an average of 0.6764 for base v3.0, from 0.3376 on sst5 to 0.9474 on snips. It reports no calibration measure (source: \u003chttps://huggingface.co/knowledgator/gliclass-base-v3.0\u003e)\n- The Ray Serve deployment answers `POST /gliclass` with a JSON array of `{label, score}` and processes one text a request (source: \u003chttps://docs.knowledgator.com/docs/frameworks/gliclass/serving/\u003e)\n- The Tests workflow failed on its last three runs on main, all on 24 September 2026, after five passing runs from 19 May to 21 July (source: \u003chttps://github.com/Knowledgator/GLiClass/actions/workflows/tests.yml\u003e)\n- Issue #47 of 20 September 2026 asks for a TypeSafe-compatible endpoint and was open on 8 October (source: \u003chttps://github.com/Knowledgator/GLiClass/issues\u003e)\n- No ONNX file is in the base v3.0 repository and the library has no export command. Issue #8 asking for an ONNX build has been open since 22 September 2024\n\n## Compare\n\n- [Celeris-1 Decision vs GLiClass](https://www.anchorterminal.com/compare/celeris-1-decision-vs-gliclass.md): D 49.3 vs D 49.9\n- [Clef vs GLiClass](https://www.anchorterminal.com/compare/cloudflare-clef-vs-gliclass.md): B 66.1 vs D 49.9\n- [Laya vs GLiClass](https://www.anchorterminal.com/compare/convai-laya-vs-gliclass.md): B 69.2 vs D 49.9\n- [Decider vs GLiClass](https://www.anchorterminal.com/compare/decider-vs-gliclass.md): B 69.5 vs D 49.9\n- [GLiClass vs Kev](https://www.anchorterminal.com/compare/gliclass-vs-jaredpalmer-kev.md): D 49.9 vs B 67.4\n- [GLiClass vs Liquid d1](https://www.anchorterminal.com/compare/gliclass-vs-liquid-d1.md): D 49.9 vs E 43.5\n- [GLiClass vs OpenAI Decisions API](https://www.anchorterminal.com/compare/gliclass-vs-openai-decisions-api.md): D 49.9 vs BB 71.5\n- [GLiClass vs Strands Decider 2B](https://www.anchorterminal.com/compare/gliclass-vs-strands-decider.md): D 49.9 vs C 61.3\n- [GLiClass vs Jev](https://www.anchorterminal.com/compare/gliclass-vs-typesafe-jev.md): D 49.9 vs B 62.1\n- [GLiClass vs Vela 2.0](https://www.anchorterminal.com/compare/gliclass-vs-vela.md): D 49.9 vs B 66.5\n\n## Verify this listing\n\nFor the vendor. The badge or a plain link to this page verifies the listing, from a page on knowledgator.com or one of its subdomains, or the README of github.com/Knowledgator/GLiClass. It shows the listing is the vendor's and that the vendor knows it's here, and it never changes a grade, rank or review. The vendor sends the page's address to `POST https://www.anchorterminal.com/api/v1/verify` as `{\"slug\": \"gliclass\", \"url\": \"…\"}`, or calls the `verify_listing` tool at https://www.anchorterminal.com/mcp. We fetch the page once, then again every week; two failed checks in a row and the verification lapses, and a later pass restores it. What we check: https://www.anchorterminal.com/builders/index.md#verify\n\nHTML badge:\n\n```html\n\u003ca href=\"https://www.anchorterminal.com/tools/gliclass\"\u003e\u003cimg src=\"https://www.anchorterminal.com/badges/gliclass.svg\" alt=\"GLiClass on Anchor Terminal\" height=\"20\"\u003e\u003c/a\u003e\n```\n\nMarkdown badge, for a README:\n\n```markdown\n[![GLiClass on Anchor Terminal](https://www.anchorterminal.com/badges/gliclass.svg)](https://www.anchorterminal.com/tools/gliclass)\n```\n\nPlain link:\n\n```html\n\u003ca href=\"https://www.anchorterminal.com/tools/gliclass\"\u003eGLiClass on Anchor Terminal\u003c/a\u003e\n```\n\n## Share this listing\n\nFor the vendor. Sharing assets for social media, two PNGs of 1200 × 630 that say GLiClass is listed on Anchor Terminal, with the vendor's logo and this page's address and no grade or score.\n\n- Dark: https://www.anchorterminal.com/assets/share/gliclass-dark.png\n- Light: https://www.anchorterminal.com/assets/share/gliclass-light.png\n", "meta": { "attribution": "Anchor Terminal (https://www.anchorterminal.com)", "docs": "https://www.anchorterminal.com/docs/", "generatedAt": "2026-10-09", "license": "CC-BY-4.0", "method": "https://www.anchorterminal.com/benchmark/", "methodology": "0.4", "openapi": "https://www.anchorterminal.com/openapi.json", "preview": false, "run": "2026-10-01", "runLabel": "October 2026 research run" }, "page": { "breadcrumbs": [ { "name": "Home", "url": "https://www.anchorterminal.com/" }, { "name": "Terminal", "url": "https://www.anchorterminal.com/tools/" }, { "name": "Decision models", "url": "https://www.anchorterminal.com/categories/decision-models" }, { "name": "GLiClass", "url": "" } ], "description": "GLiClass is an open-source Python library and family of open-weight zero-shot text classifiers from Knowledgator. It scores every candidate label in one forward pass and runs locally through a pipeline or a Ray Serve endpoint.", "facts": [ "rank #697 of 842", "None auth", "0 desk reviews" ], "h1": "GLiClass", "image": "https://www.anchorterminal.com/assets/og/tools-gliclass.png", "path": "/tools/gliclass", "published": "2026-10-01", "section": "tools", "title": "GLiClass review for AI agents, grade D (49.9/100) | Anchor Terminal", "toc": null, "updated": "2026-10-09", "url": "https://www.anchorterminal.com/tools/gliclass" }, "tokens": { "markdown": 6200, "slim": 1580 }, "version": 1 }