# Drex Max v1 > Drex Max v1 is Nace.AI's open-weight 27B decision model, a fine-tune of Qwen3.8-27B that takes text, JSON and images with typed questions and returns a probability for every option without generating text. - Canonical: https://www.anchorterminal.com/tools/drex-max-v1 - Pricing: https://www.anchorterminal.com/tools/drex-max-v1#pricing - Reviews: https://www.anchorterminal.com/tools/drex-max-v1#reviews - Markdown: https://www.anchorterminal.com/tools/drex-max-v1.md (~8,200 tokens) - Slim: https://www.anchorterminal.com/tools/drex-max-v1.min.md (~2,430 tokens, same facts, less prose, for token-sensitive contexts) - JSON: https://www.anchorterminal.com/tools/drex-max-v1.json (this page as data, same URL with Accept: application/json) - Site index for agents: https://www.anchorterminal.com/llms.txt (full text: https://www.anchorterminal.com/llms-full.txt) - API: https://www.anchorterminal.com/api/v1/index.json - MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key) - Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data) - Updated: 2026-10-30 ## Overview **Grade C · 55.5/100 · rank #90 of 120 open-weight models · #17 in Decision models · not agent-ready · confidence low** The same model, listed separately and graded on its own: [Drex 1.5](https://www.anchorterminal.com/tools/nace-drex.md) (Model API). Trained by Nace.AI, in the Drex family, with [Drex v1.5](https://www.anchorterminal.com/tools/nace-drex.md), [Drex DLM](https://huggingface.co/nace-ai/drex-dlm), [Drex v2.0](https://huggingface.co/nace-ai/drex-v2.0). Every model from this lab and its older versions: https://www.anchorterminal.com/open-weights/#nace. ## Assessment A 27B cousin for typed decisions over text and screenshots, several questions per request with a probability for each option. Images need an NVIDIA GPU and Nace's loader; on Apple silicon only the text backbone runs, from 24 GB. The licence isn't open source, with free commercial use only under $1 million in revenue and funding. ## Facts | Field | Value | | --- | --- | | Vendor | Nace AI, Inc. (Nace.AI) (https://www.nace.ai) | | Kind | Open-weight model | | Category | Decision models (https://www.anchorterminal.com/categories/decision-models) | | Auth | None · The weights download from Hugging Face with no account, token or terms to accept. The repository and both quantised builds are not gated (checked 2026-10-30). The licence says downloading binds you to its terms. Nace's hosted API for the same model needs a nace_sk_ key. | | Pricing | Free (Free) · Free to download. Commercial use by anyone above $1 million in revenue or funding, or with a product that competes with Nace's, needs a separate licence from Nace.AI. Running it costs the hardware, from 14.5 GB of files (affine 4-bit, text only) to 52.2 GB (BF16). Nace's API serves it as drex-max at $0.03 per million input tokens, with output not billed (checked 2026-10-30). | | x402 | No · Weights you download and run yourself, so there is nothing to pay per call and no payment protocol (checked 2026-10-30). | | Licence | Nace.AI Open RAIL-M License (modified), version 1.0, October 2026, in the repository's LICENSE. Not an open-source licence. Use is free only for individuals and organisations under $1 million in revenue in the prior 12 months and under $1 million in total funding (personal and research use excepted), never by anyone with a product that competes with Nace's, decision models included, and under use restrictions that bind the model, its derivatives and its outputs. Built on Qwen/Qwen3.8-27B (Apache-2.0), whose licence text isn't carried in the repository | | Packages | huggingface: `nace-ai/drex-max-v1` | | Source | https://huggingface.co/nace-ai/drex-max-v1 | | Docs | https://github.com/nace-ai/drex-decision-models/blob/main/models/drex-max-v1 | | llms.txt | not found | | Last release | 2026-10-30 | | Surface graded | The open weights at huggingface.co/nace-ai/drex-max-v1 and the publisher's Q8_0 and affine 4-bit builds, read from the cards and their history, the files and their metadata, config.json, decision_config.json, the loader in code/drex_max, inference.py, the licence, Nace's GitHub docs and MLX script, Nace's blog, product page and hosted docs, the Jev Decision Index Space's data and Artificial Analysis's index on 30 October 2026. Nothing was downloaded or run. Nace's hosted API is the nace-drex listing | | Built on | Qwen/Qwen3.8-27B (Apache-2.0) at revision 1d4bf0f, named in decision_config.json and the loader, and in the card's metadata until 30 October 2026. A decision fine-tune of the backbone with a new 255-way readout head. Nace doesn't say how it was trained | | Relation to Drex 1.5 and the hosted API | The same publisher's larger, multimodal line. nace-drex is graded as Nace's hosted Drex API at console.nace.ai/v1/systemone, which today serves drex-max (this model, released 13 October 2026), drex-max-v1.1, drex-v2.0 (the default) and drex-v1.5, and forwards the retired drex-v1.0 to drex-max. Whether the hosted drex-max is byte for byte the Hub checkpoint isn't stated | | Model | 26,085,330,160 parameters in BF16 by the Hub's count (the card says 27B; the base counts 27.78B with its language-model head), 64 layers with 48 Gated DeltaNet layers that stay causal and 16 full-attention layers made non-causal, the Qwen3.5 vision tower, and a 255 by 5,120 readout over the last hidden state with a temperature of 1.009 | | Files | 11 safetensors shards (52.17 GB), readout.safetensors, decision_config.json, config, tokenizer, processor config and the base's chat_template.jinja, the loader package code/drex_max, inference.py and requirements.txt. No pickle files. The affine 4-bit build adds pointer.bin, the readout in Nace's raw KPS1 format, not a pickle. The Hub lists a SHA-256 for every LFS file, and the builds' cards list theirs | | Licence | Nace.AI Open RAIL-M (modified) 1.0. Royalty-free copyright and patent grants, subject to Attachment A, which excludes users above $1 million in revenue or funding (personal and research use excepted) and anyone with a product competing with Nace's, decision models and document intelligence included, and bars fully automated decisions with legal or similar effects without human review, discrimination, unsupervised medical, legal or financial advice, military, weapons and mass-surveillance use, fraud and removing the licence. Derivatives, distillations and outputs carry the restrictions. No clause lets Nace change the terms. The Apache-2.0 base's licence text isn't in the repository | | Runtimes | Nace's loader on released Transformers 5.17.0 (pinned in requirements.txt) runs it as trained with images, on CUDA. The text backbone also runs on Nace's llama.cpp and Ollama forks (branch drex-max-v1), Kapitsa 0.2.25 or later (Nace's Apple silicon engine, Apache-2.0) and Nace's drex_max_mlx.py on MLX. No released llama.cpp, Ollama, vLLM, SGLang or LM Studio route, and no Ollama library entry | | Context | 262,144 tokens in config.json (rope default) and the docs, the base's native window. The loader defaults to 16,384 tokens per request (inference.py in the repository defaults to 8,192) and accepts up to 262,144 with DREX_MAX_CONTEXT or --max-length. The fine-tune's training length isn't stated | | Inputs and outputs | A state (text or JSON), optional images (file paths or data URLs, up to 4,194,304 pixels each), and named questions, each choice (1 to 255 options with descriptions), noul (yes or no, criteria optional) or score (ordered levels). Each question is scored as one pass over the state and returns probabilities, with choice and confidence, P(true), or an expected score and confidence. The system prompt tells the model to treat the state as data. No text generation and no thinking | | Hardware | The card's tested hardware is an NVIDIA H100 80 GB in BF16 (48.6 GiB loaded), an NVIDIA L40S 48 GB with the backbone in 8-bit (25.9 GiB), and an Apple M5 Pro with 48 GB for the Q8_0 GGUF on Metal. The affine 4-bit build is stated for Macs with 24 GB or more | | Training data | Not described. The card, the GitHub page and the blog say nothing about the decision fine-tune's data, size or cut-off. The nace-ai datasets for this model hold Decision Index results, not training data | | Capabilities | inference.decision, model.open-weights | | Tags | open-weights, model, decision-model, full-fine-tune, calibrated-probabilities, vision, open-rail-m, safetensors, gguf, self-hosted, apple-silicon | | JSON | https://www.anchorterminal.com/api/v1/tools/drex-max-v1.json | ## Score breakdown (methodology v0.7, October 2026 research run) Assessed 2026-10-30 from public evidence against the published checklist (https://www.anchorterminal.com/benchmark/#checklist). Confidence: low. 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. | Category | Weight | This run | Score (0–100) | Points | | --- | --- | --- | --- | --- | | Capability | 35% | pending | pending | n/a | | Reach | 25% | 38.5 | 65 | 25.0 | | Risk | 20% | 30.8 | 45 | 13.8 | | Agent interface | 10% | 15.4 | 80 | 12.3 | | Documentation and upkeep | 10% | 15.4 | 41 | 6.3 | | Negative events | up to −15 | up to −15 | 19 October 2026, still on the page on 30 October. Nace's blog says Drex Max is 'First among decision models on the public Decision Index 0.4.0'. The index (multimodalart/jev-decision-index, data generated 29 October 2026) ranks it 23rd of 247 by public score at 67.52, behind mammoth-1 (75.34) among others, and 24th of 247 by the Full score it ranks by (66.6). Its data of 16 October, before the blog, already had StartLux-Decision v1.1 27B (72.88), KnowLine 35B-A3B Gen1 (69.49) and Perplexity Decider v1.2 (67.74) above 67.52 on the public score. The Vision board ranks it 11th of 62. Sources: https://www.nace.ai/blog/introducing-drex-max and https://huggingface.co/spaces/multimodalart/jev-decision-index/blob/main/data/index.json (-2) | -2 | | **Total** | | | | **55.5 → C** | ### Why each score - Capability: 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. - Reach 65 (checked 2026-10-30): Memory 25 of 30, Nace's affine 4-bit build is 14.5 GB, 17.4 GB with a fifth added, so it fits 24 GB, though it reads text only. Runtimes 5, released Transformers runs it with Nace's loader in the repository, while llama.cpp and Ollama need Nace's forks. Builds 15, a Q8_0 GGUF and a 4-bit build from Nace. Terms 10, a custom Open RAIL-M licence with free commercial use under $1 million in revenue and funding and a use policy. Download 10, not gated. Hosted 5, Nace's own API serves it as drex-max. - Risk 45 (checked 2026-10-30): File safety 15 of 25, safetensors and GGUF only, but the documented path runs the repository's loader code. Provenance 15, the organisation nace.ai links to, with a checksum on every file. Safety evaluation 0, none for the model. Agent guidance 5, the licence's human-review terms and a system prompt in code, with no documented injection guidance. Licence risk 5, field-of-use limits and a revenue and funding threshold. Security programme 5, a valid security.txt but no public advisories. - Agent interface 80 (checked 2026-10-30): Read as a decision model, by what it takes and returns. Chat template 10 of 20, the base's template is there but the decision prompt is built in code. Tool-calling format 25, the card and request-format doc give the request, the answers and the calls. Trained tool calling 20, several questions per request. Structured output 15, typed JSON answers shown for every type. Reasoning control 0, it doesn't think. Context 10, a 262,144-token window with no scaling. - Documentation and upkeep 41 (checked 2026-10-30): Model card 11 of 20, four of the seven items. Training data 0, nothing on the fine-tune's data. Last change 20, the card and a new build on 30 October. Responsiveness 0, no discussions yet. Version policy 5, a first release. File health 5, the card dropped its base while the files still name it, and inference.py's default length differs from the card's. Fix list for a coding agent, everything this grade says the listing lacks, the biggest gain first (15 items): https://www.anchorterminal.com/fixes/drex-max-v1.md (JSON https://www.anchorterminal.com/fixes/drex-max-v1.json) ### What we couldn't check - Whether Nace's hosted drex-max, which Artificial Analysis benchmarked, is the same checkpoint as the Hub's files isn't stated. The release dates match (13 October 2026) and the Decision Index's served parameter count equals the Hub's - The runtimes line credits Transformers 5 because Nace's own loader runs on released, unpatched Transformers. Read like mammoth-1 and joey-1, where only a runtime that applies the non-causal mask itself counts, it's 0 and Reach is 65 - The fine-tune's training length isn't stated. Context is credited at 262,144 from config.json and the docs; if training ran shorter, the line falls - unchecked: issues and pull requests on github.com/nace-ai/drex-decision-models. GitHub's site and API weren't reachable from this session, only raw.githubusercontent.com - The card named Qwen/Qwen3.8-27B as its base until 30 October 2026 and now names none, while the licence chain still runs through Apache-2.0. Recorded in file health, not deducted - Whether the repository was public before 14 October 2026, when its first licence (non-commercial research only) was replaced, couldn't be established from the Hub API. Read as replaced before release ### Sources - model card, read 30 October 2026: (seen 2026-10-30) - Hub API, card data, gating, downloads, parameter count, providers: (seen 2026-10-30) - file tree with sizes and SHA-256 for LFS files: (seen 2026-10-30) - commit history, the licence change of 14 October and the base removal of 30 October: (seen 2026-10-30) - the card before base_model was removed: (seen 2026-10-30) - the first card, built on Qwen3.8-27B, private flag: (seen 2026-10-30) - Nace.AI Open RAIL-M (modified) 1.0, read in full: (seen 2026-10-30) - architecture, layer types, positions: (seen 2026-10-30) - base model and revision, option codes, temperature, non-causal attention: (seen 2026-10-30) - loader, attention hook, readout, decision prompt: (seen 2026-10-30) - repository script, default length: (seen 2026-10-30) - Transformers 5.17.0 pinned: (seen 2026-10-30) - Q8_0 GGUF card, size, hash, runners: (seen 2026-10-30) - affine 4-bit card, files, hashes, measurements: (seen 2026-10-30) - affine 4-bit file sizes: (seen 2026-10-30) - base revision, parameter count, Apache-2.0, chat template blob: (seen 2026-10-30) - the publisher's models and builds: (seen 2026-10-30) - organisation members, the committing accounts among them: (seen 2026-10-30) - Nace's code and docs, release date, runners, read through raw.githubusercontent.com: (seen 2026-10-30) - request format: (seen 2026-10-30) - context limits per runner: (seen 2026-10-30) - MLX runner and the pointer.bin format: (seen 2026-10-30) - Kapitsa, Nace's Apple silicon engine, Apache-2.0: (seen 2026-10-30) - launch blog of 19 October 2026, the 'First among decision models' claim, benchmark table: (seen 2026-10-30) - product page, Drex Max v1.1 figures, links to the Hub and GitHub: (seen 2026-10-30) - hosted models, drex-max and drex-max-v1.1, prices, limits: (seen 2026-10-30) - hosted Drex Max guide: (seen 2026-10-30) - security.txt, valid until 30 October 2027: (seen 2026-10-30) - Jev Decision Index 0.4.1 data generated 29 October 2026, Drex Max's entry and ranks: (seen 2026-10-30) - Vision board data generated 29 October 2026: (seen 2026-10-30) - index data of 16 October 2026, public scores before the blog: (seen 2026-10-30) - index README, ranked by Full score, not affiliated with TypeSafe AI: (seen 2026-10-30) - Artificial Analysis Decision Index, Drex Max 63.74: (seen 2026-10-30) ## Score history Every move of a score since the listing was graded, newest first, with its cause. A score can't move without an entry here. JSON: https://www.anchorterminal.com/api/v1/tools/drex-max-v1/history.json - 2026-10-30: graded 55.5 C (methodology 0.7) ## Who's behind it (provenance 58/100, checked 2026-10-30) | Check | Finding | Points | | --- | --- | --- | | Legal entity named | Nace AI, Inc. | 20/20 | | Domain age | nace.ai, no registry record we could read | 0/15 | | Endpoint on the vendor's domain | no hosted endpoint | n/a | | Terms of service | nothing hosted, so the Nace.AI Open RAIL-M License (modified), version 1.0, October 2026, in the repository's LICENSE. Not an open-source licence. Use is free only for individuals and organisations under $1 million in revenue in the prior 12 months and under $1 million in total funding (personal and research use excepted), never by anyone with a product that competes with Nace's, decision models included, and under use restrictions that bind the model, its derivatives and its outputs. Built on Qwen/Qwen3.8-27B (Apache-2.0), whose licence text isn't carried in the repository licence stands in | 10/10 | | Privacy policy | nothing hosted, not scored | n/a | | Status page | not found | 0/10 | | Changelog | not found | 0/10 | | security.txt | valid | 10/10 | | People named | 3 named with their role, none with a public profile | 7/10 | | Support route | help pages, a community | 5/5 | The legal entity is from the nace-drex listing's check of Nace's terms on 10 October 2026. The nace-ai Hugging Face organisation is unverified, with 19 members, among them the two accounts that committed to this repository. nace.ai/drex links to huggingface.co/nace-ai and to github.com/nace-ai, whose drex-decision-models repository links these weights. Nothing is hosted under this listing, so the licence stands in for terms of service and no privacy policy is scored. security.txt at www.nace.ai is valid until 30 October 2027 and points to trust.nace.ai for contact and policy. ### Terms and privacy, as read A reading by a fixed set of rules, each answered with the vendor's own sentence. Not legal advice. **Terms of service**. Nothing is hosted by the vendor, so there are no terms of service to read. The Nace.AI Open RAIL-M License (modified), version 1.0, October 2026, in the repository's LICENSE. Not an open-source licence. Use is free only for individuals and organisations under $1 million in revenue in the prior 12 months and under $1 million in total funding (personal and research use excepted), never by anyone with a product that competes with Nace's, decision models included, and under use restrictions that bind the model, its derivatives and its outputs. Built on Qwen/Qwen3.8-27B (Apache-2.0), whose licence text isn't carried in the repository licence stands in and the check scores in full. **Privacy policy**. Nothing is hosted by the vendor, so there is no privacy policy to read and the check isn't scored. ## Humans behind the project (researched 2026-10-30 from public pages, not part of the grade) ### Key people - Dos Bahá, Founder and CEO - Zhanibek Datbayev, Co-founder and CTO - Sudha Valluru, Co-founder and COO - Note: started from data/humans/nace-drex.json (nace ai, inc., checked 28 october 2026). the three founders and their roles are from nace.ai's press page as read there and weren't re-read today. the drex max repositories are committed by two members of the nace-ai hugging face organisation, whose roles aren't stated on a nace page, so they aren't listed. ownership and investors are left out so land.py copies them from the nace-drex file; nace.ai's sitemap, read on 30 october 2026, lists no blog post after the seed announcement about funding, and its newest entries date from 26 october. the support email address in the nace-drex file isn't copied here, since this run's rules copy no email addresses into files. - Nace AI, Inc. (Nace.AI): LinkedIn https://www.linkedin.com/company/nace-ai/ X https://x.com/NaceAI GitHub https://github.com/nace-ai ### Ownership and investors - Raised: $21.5M in total, as of May 2026 (source https://www.nace.ai/blog/announcing-our-seed-investment) - Backed by: Walden Catalyst Ventures, General Catalyst ### Getting support - Discord: https://discord.com/invite/nace-ai - Help: https://trust.nace.ai - Sources: https://www.nace.ai/press, https://www.nace.ai/, https://www.nace.ai/about, https://www.nace.ai/drex, https://www.nace.ai/blog/introducing-drex-max, https://www.nace.ai/.well-known/security.txt, https://nace.ai/sitemap.xml ## Probe metrics Not 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. No hosted version of these weights has a price on OpenRouter, so what a task costs is your own hardware and power. ## Strengths - Probabilities for choice, yes or no and score questions over text, JSON and images, with nothing generated - Full score 66.6 on the Jev Decision Index 0.4.1, 24th of 247, and 11th of 62 on its Vision board, from the index's private tests - Nace publishes a Q8_0 GGUF and an affine 4-bit build, the smaller at 14.5 GB for Macs with 24 GB - Not gated, and Nace's own API serves it as drex-max at $0.03 per million input tokens - 262,144-token window in config.json and the docs, with a 16,384-token default you can raise ## Weaknesses - The licence bars commercial use above $1 million in revenue or funding and by anyone with a competing product, and restricts the outputs too - Every path that runs it as trained is Nace's own code or forks. Stock llama.cpp, Ollama, vLLM and Transformers run it causally - Images need the Python loader on a CUDA GPU; the GGUF and 4-bit builds read text only - The card gives no limitations, languages, cut-off or training data, and dropped its Qwen3.8-27B base on 30 October 2026 - No safety results and no prompt-injection guidance on the card ## Before you call it (notes for agents) 1. Clone github.com/nace-ai/drex-decision-models, download nace-ai/drex-max-v1 and run `python -m drex_max --model weights/drex-max-v1 --request examples/request.json` on an NVIDIA GPU 2. Plan for about 49 GB of GPU memory in BF16, or about 26 GB with --8bit on CUDA, before any long context 3. Don't load it with stock Transformers, vLLM or llama.cpp. Its full-attention layers must run non-causally with readout.safetensors applied, which only Nace's loader and forks do 4. On a Mac with 24 GB or more, run `kapitsa-systemone --model ./drex-max-v1-affine-q4 --port 8000 --context 2048` for text-only decisions at POST /v1/systemone 5. Raise the 16,384-token default with DREX_MAX_CONTEXT=N or --max-length N, up to 262,144, and expect more memory 6. Check the licence first. Above $1 million in revenue or funding, or with a competing product, you need Nace's commercial licence ## Get started Install: ```bash git clone https://github.com/nace-ai/drex-decision-models && cd drex-decision-models hf download nace-ai/drex-max-v1 --local-dir weights/drex-max-v1 pip install -r models/drex-max-v1/requirements.txt python -m drex_max --model weights/drex-max-v1 --request examples/request.json # NVIDIA GPU, Python 3.12+; add --8bit for about 26 GB ``` First request: ```bash kapitsa-systemone --model ./drex-max-v1-affine-q4 --port 8000 --context 2048 # Apple silicon, text only, after brew install nace-ai/tap/kapitsa and hf download nace-ai/drex-max-v1-affine-q4 ``` ## Similar tools Ranked by shared capabilities, then score. Same-category tools with no shared capability key are listed last. | Tool | Grade | Score | Rank | Shared capabilities | x402 | Markdown | | --- | --- | --- | --- | --- | --- | --- | | JEV-27B-VL | C | 59.5 | open-weight #67 | inference.decision, model.open-weights | no | https://www.anchorterminal.com/tools/jev-27b-vl.md | | joey-1 | C | 57.1 | open-weight #80 | inference.decision, model.open-weights | no | https://www.anchorterminal.com/tools/joey-1.md | | mammoth-1 | D | 47.7 | open-weight #115 | inference.decision, model.open-weights | no | https://www.anchorterminal.com/tools/mammoth-1.md | | elephant-1 | D | 47.6 | open-weight #116 | inference.decision, model.open-weights | no | https://www.anchorterminal.com/tools/elephant-1.md | | Qwen3-Embedding-8B | BB | 74.7 | open-weight #1 | model.open-weights | no | https://www.anchorterminal.com/tools/qwen3-embedding-8b.md | | Gemma 4 31B IT | BB | 74.1 | open-weight #2 | model.open-weights | no | https://www.anchorterminal.com/tools/gemma-4-31b-it.md | ## Panel reviews (0) Reviewed by the Anchor panel (https://www.anchorterminal.com/reviewers/index.md): . Desk 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 ## Notable - Built on Qwen/Qwen3.8-27B. decision_config.json names it as base_model at revision 1d4bf0f, which is the Qwen repository's current commit, the loader's docstring says 'Qwen3.8-27B hybrid', and chat_template.jinja is the same blob as Qwen's. The card declared the base in its metadata until a commit on 30 October 2026 at 11:37 UTC removed it, and now calls the model 'a dense 27B model' with no base named (source: , ) - Nace's blog of 19 October 2026 calls Drex Max 'First among decision models on the public Decision Index 0.4.0'. The index, a community Space by multimodalart that says it isn't affiliated with TypeSafe AI, ranks Drex Max 23rd of 247 by public score (67.52) and 24th of 247 by the Full score it ranks by (66.6, tie group 5) in data generated 29 October. On its 16 October data, before the blog, three models already had higher public scores (source: , ) - On the index's Vision board, which ranks image-capable decision models, Drex Max is 11th of 62 by Full score (71.25) and 13th by public score (74.09). WeMM-JEV-27B leads at 73.81 (source: ) - Artificial Analysis's Decision Index, run by Artificial Analysis on Nace's hosted Drex Max, gives it 63.74 on 30 October 2026, fourth of the 18 rows in its default view, behind d1-ultra (65.25), SubQ Snap 1.1 (65.00) and Elephant 9B (64.69). The card's '64, fourth of 18' matches (source: ) - On Apple silicon only the text backbone runs. The Q8_0 GGUF runs on Nace's own forks of llama.cpp and Ollama (branch drex-max-v1) on Metal or the CPU and needs 48 GB, and the affine 4-bit build runs on Kapitsa, Nace's own Apple silicon engine, or Nace's MLX script on Macs with 24 GB or more. Images need the Python loader on an NVIDIA GPU. The card's tested Mac is an M5 Pro with 48 GB, and no M4 run is stated (source: ) - The 16 full-attention layers are non-causal and a 255-way readout in readout.safetensors replaces the language-model head. Nace's loader applies both on released Transformers 5.17.0 through a forward hook, and the card says a plain causal chat run doesn't reproduce its scores (source: ) - Nace's hosted Drex API, listed as nace-drex, serves this model as drex-max (released 13 October 2026) and a newer drex-max-v1.1 (28 October) whose weights aren't published, both at $0.03 per million input tokens. The 70.8 Decision Index figure on nace.ai/drex is for v1.1, not these weights (source: , ) - 52 downloads and 10 likes on 30 October 2026. The shard SHA-256 values are the same as at the first upload on 13 October. The licence file was replaced in the release commit of 14 October and hasn't changed since. No copy or quantisation by anyone else was found on the Hub (source: ) - Best decision models for AI agents: https://www.anchorterminal.com/best/decision-models/index.md - All 200 decisions comparisons: https://www.anchorterminal.com/compare/decision-models/index.md ## Compare - [Drex Max v1 vs elephant-1](https://www.anchorterminal.com/compare/drex-max-v1-vs-elephant-1.md): C 55.5 vs D 47.6 - [Drex Max v1 vs JEV-27B-VL](https://www.anchorterminal.com/compare/drex-max-v1-vs-jev-27b-vl.md): C 55.5 vs C 59.5 - [Drex Max v1 vs joey-1](https://www.anchorterminal.com/compare/drex-max-v1-vs-joey-1.md): C 55.5 vs C 57.1 - [Drex Max v1 vs mammoth-1](https://www.anchorterminal.com/compare/drex-max-v1-vs-mammoth-1.md): C 55.5 vs D 47.7 ## Verify this listing For the vendor. The badge or a plain link to this page verifies the listing, from a page on nace.ai or one of its subdomains. 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": "drex-max-v1", "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 HTML badge: ```html Drex Max v1 on Anchor Terminal ``` Markdown badge, for a README: ```markdown [![Drex Max v1 on Anchor Terminal](https://www.anchorterminal.com/badges/drex-max-v1.svg)](https://www.anchorterminal.com/tools/drex-max-v1) ``` Plain link: ```html Drex Max v1 on Anchor Terminal ``` ## Share this listing For the vendor. Sharing assets for social media, two PNGs of 1200 × 630 that say Drex Max v1 is listed on Anchor Terminal, with the vendor's logo and this page's address and no grade or score. - Dark: https://www.anchorterminal.com/assets/share/drex-max-v1-dark.png - Light: https://www.anchorterminal.com/assets/share/drex-max-v1-light.png - Create a video and download its frames: https://www.anchorterminal.com/assets/listing-video/index.html?listing=drex-max-v1 --- Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp