Agents-A1-4B

# Fix list: Agents-A1-4B

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/agents-a1-4b, the October 2026 research run, assessed 21 October 2026. Grade C, 61 out of 100.

This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public.

For a coding agent working on Agents-A1-4B: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published.

## 1. Risk, 55 out of 100, up to 13.8 more on the total

Why it scored 55: File safety 25, safetensors and GGUF only and nothing from the repository runs. Provenance 15, the files are in the InternScience organisation its GitHub repository links to, with a SHA-256 for every weight file. Safety evaluation 0 and agent guidance 0, nothing published. Licence risk 15, Apache 2.0 through the chain from Qwen3.5-4B with no conditions added. Security programme 0, no policy, contact or security.txt.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-risk):

What could go wrong when you run it, and what the publisher has done about it. A higher score means less risk.

- 0 to 25, file safety. 25 for weights in safetensors or GGUF only, 15 when pickle files (.bin, .pt, .pth) sit beside them, 0 for pickle only. 10 off when the documented way to load it or to build its prompts runs code from the repository (`trust_remote_code`, or the publisher's scripts in place of a template).
- 0 to 15, provenance. 10 when the files come from the publisher's own organisation, the one its site or docs link to, and 5 for a checksum on every file.
- 0 to 20, safety evaluation. 20 when results are published and someone other than the publisher ran them or can rerun them, 10 for the publisher's own results, 0 for none.
- 0 to 15, guidance for agent use. 15 for documented prompt-injection guidance or an instruction hierarchy the model was trained to follow, 10 for a statement of what its trained refusals cover and the safeguards to add around it, 5 for general cautions, 0 for none.
- 0 to 15, licence risk. 15 to start, 5 off for each of these, a use policy or field-of-use limits beyond the law, terms the licensor can change or revoke, a revenue or user threshold, and a licence that changed without a dated note or whose link is broken. Not below 0.
- 0 to 10, a security programme. 5 for a way to report a flaw (security.txt, a disclosure policy or a security contact), 5 for advisories handled in public.

When our reference-machine runs start, measured robustness (prompt injection with AgentDojo, over-refusal with XSTest) joins Risk.

## 2. Documentation and upkeep, 43 out of 100, up to 8.8 more on the total

Why it scored 43: Model card 8, it covers intended use, parameters and context but not the architecture, limitations, languages or a knowledge cut-off. Training data 5, a general description with no cut-off. Last change 10, the files went up on 14 July 2026, 99 days ago. Responsiveness 10, InternScience answered one of the three discussions. Version policy 5, a first release. File health 5, the files agree on the essentials, but the config declares a prediction layer that isn't shipped and the template carries a dated default prompt.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-upkeep):

- 0 to 20, the model card. Intended use, limitations, architecture, parameters, context length, languages and knowledge cut-off. 20 for all seven, 3 off for each that's missing.
- 0 to 20, training data. 20 for the data published or listed with sources and sizes, 15 for a data card that names the sources or the mix, the cut-off and the filtering, 10 for a general description with a cut-off, 5 without one, 0 for none.
- 0 to 20, the last change to the files, the template or the reference code. 20 within 30 days, 15 within 90, 10 within 180, 5 within a year, 0 older.
- 0 to 20, responsiveness. The publisher's replies to the 40 most recent questions and reports in the repository's discussions or issues. 20 for most, 10 for some, 5 for a few, 0 for none.
- 0 to 10, version policy. 10 for earlier versions kept in their own repositories with a dated note when one is replaced, 5 for earlier versions kept without notes or a first release, 0 when files were replaced in place without a note.
- 0 to 10, file health. 10 when config, tokenizer, template and card agree and no report of a broken file waits without a reply, 5 when one does, 0 when the files contradict the card.

## 3. Agent interface, 70 out of 100, up to 4.6 more on the total

Why it scored 70: Chat template 20, a Jinja file in the repository. Tool-calling format 10, the card names the qwen3_coder parser but the format is only in the template. Trained tool calling 20, the card says it was trained for tool use and the template renders several calls in one turn. Structured output 0. Reasoning control 10, the template turns thinking off per request and the card's reasoning parser returns it apart from the answer. Context 10, 262,144 tokens.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-interface):

What a model and a developer need to talk to it. Speech, image and other models with no chat interface read the tool-calling and structured-output lines for what the model takes and returns (voices, timestamps, streaming, typed outputs), and the note says how.

- 0 to 20, the chat template. 20 for a template in the repository (Jinja, in the tokenizer config or its own file), 10 when the prompt format is only given as code you run.
- 0 to 25, the tool-calling format. 25 when the publisher documents how tools are declared and how calls come back, with the parser to use in a named runtime, 10 when only the template or a runtime's own recipe says it.
- 0 to 20, tool calling the model was trained for. 20 with several calls in one turn, 10 for one call at a time.
- 0 to 15, structured output. 15 when JSON or JSON Schema output is documented for the model with how to ask for it, 5 when the card claims it without saying how.
- 10, reasoning control. Thinking that can be turned off or budgeted per request, with where the reasoning text comes back.
- 0 to 10, context. 10 for 128K tokens or more trained, 5 for 32K to 127K. Context reached only by scaling the position encoding counts one band down.

## 4. Reach, 90 out of 100, up to 3.8 more on the total

Why it scored 90: Memory 30, InternScience's Q4_K_M GGUF with its vision projector is 3.4 GB and needs about 4.1 GB with the fifth added. Runtimes 20, the Qwen3.5 architecture runs in released vLLM, SGLang, Transformers and llama.cpp, while Ollama's and LM Studio's entries are the base's, not this model's. Builds 15, InternScience publishes FP8 and three GGUF precisions. Terms 15, Apache 2.0 with nothing added. Download 10, nothing is gated. Hosted 0, one inference provider serves it and the line needs two.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reach):

Whether an agent can run it, and where.

- 0 to 30, memory. The smallest build from the publisher or a runtime's own library that keeps 4-bit precision or better, at its file size plus a fifth for the runtime and a working context. 30 for 16 GB or less, 25 for 24 GB, 20 for 32 GB, 15 for 64 GB, 10 for what fits our reference machine (the GPU share of 128 GB of unified memory), 5 for 256 GB, 0 above.
- 0 to 25, runtimes. 5 for each of llama.cpp, Ollama, vLLM, SGLang, LM Studio and Hugging Face Transformers that runs the architecture in a released version, from the runtime's own code, library or documentation. A publisher's fork or an unreleased branch doesn't count. 25 at most.
- 0 to 15, builds from the publisher. 15 for quantised builds at two precisions or more (FP8, 4-bit, GGUF), 10 for one, 0 when only other people's quantisations exist.
- 0 to 15, terms of use. 15 for an OSI-approved licence with nothing added, 10 for a custom licence that lets anyone use it commercially for free below a threshold or under a use policy, 5 for research or non-commercial use only, 0 for no licence or unclear terms.
- 0 to 10, download. 10 when the files download with no account and no terms to accept, 5 for a click-through that needs an account, 0 when a person approves each request.
- 5, hosted too. The publisher's own API serves it, or at least two inference providers do, so an agent without the hardware can still call it.

When our reference-machine runs start, speed on that machine (tokens per second and time to first token at the build that fits) joins Reach.

## What we couldn't check

What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it.

- The base is inferred from the files (Qwen3.5-4B), since neither the card, the metadata nor the technical report names it for the 4B
- unchecked, GitHub issues on InternScience/Agents-A1, since the GitHub API wasn't available, so responsiveness rests on the three Hugging Face discussions
- Reasoning control is credited on the template's switch and the card's reasoning parser, as for Ornith, though the card doesn't name the switch
- Hosted would reach 5 if a second inference provider or InternScience's own API served it

## Weaknesses

- The base isn't declared, and the technical report doesn't cover the 4B
- No safety results, agent guidance or security contact
- Without a system message the template inserts its own prompt naming the model Intern-A1 with a fixed date of 2026-07-14
- Tool calling and structured output aren't documented beyond a parser name
- InternScience answered one of three discussions, and the history was squashed

## What costs an agent a turn today

The notes we give agents before they call it. Each one is a workaround an agent shouldn't need.

- Quickest start is `llama-server -hf InternScience/Agents-A1-4B-Q4_K_M-GGUF --jinja`, about 3.4 GB with the vision projector
- With vLLM use `--reasoning-parser qwen3 --enable-auto-tool-choice --tool-call-parser qwen3_coder`, as on the card, and add `--language-model-only` for text-only work
- Always send your own system message, or the template inserts InternScience's default prompt with the date fixed at 2026-07-14
- The card recommends temperature 0.85, top_p 0.95, top_k 20, min_p 0 and presence_penalty 1.1, and there is no generation_config.json to set them
- Thinking is on by default, send `chat_template_kwargs` with `enable_thinking` false for short tasks, a switch the card doesn't mention
- The config declares a multi-token prediction layer that isn't shipped, so don't turn on MTP speculative decoding
- Validate JSON output yourself, since structured output isn't documented

## What the review panel asked for

- ship a generation config
- answer discussions
- a declared base model
- a tool-calling section on the card
- a security contact
- name the base model

## When it's done

Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.