Model

METAGENE-1

METAGENE-1 is an autoregressive DNA/RNA sequence model trained on wastewater metagenomic data.

Sources (4)

metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length

Results are available for configurations using this model. Their fitted heads, extra inputs and evaluation settings are kept separate below.

1 evaluated configuration using this model

How it worksMETAGENE-1 workflow

Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation.

Sources (4)

Overview

Model type

Autoregressive metagenomic transformer

Inputs

DNA or RNA nucleotide sequences.

Outputs

Sequence generation and representations for downstream metagenomic analyses.

Access

Official downloadable model/card and usage examples: https://huggingface.co/metagene-ai/METAGENE-1

Sources (4)

limited source coverage · Automated source review, 2026-09-16. All specifications and missing details

Related configurations, pipelines and services

These configurations, services and pipelines use this model within their own configurations. Their results, where available, are not assigned to the underlying model.

METAGENE-1 Pipeline · 16 results

Use this model

How it works, versions and access

How it works

METAGENE-1 is an autoregressive DNA/RNA sequence model trained on wastewater metagenomic data. Llama-style autoregressive transformer with 32 layers, width 4,096, 32 attention heads and a 1,024-token vocabulary in the inspected configuration. The documented inputs are DNA or RNA nucleotide sequences. The output consists of sequence generation and representations for downstream metagenomic analyses.

Sources (4)

Versions and reproducibility

METAGENE-1; exact model-card and configuration revision pinned in sources. Configuration fields differ: max_position_embeddings=512 and max_sequence_length=2,048. The effective supported window remains unresolved; neither value is silently promoted to a validated inference limit.

Sources (4)

Strengths, limitations and unresolved questions

Strengths and limitations

Strengths and considerations

Its pretraining corpus includes diverse mixed-community DNA and RNA rather than a single reference genome.
Sources (4)
metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length

Limitations and conditions

The stated biosurveillance and pathogen-detection applications require their own downstream evaluation; the pretraining objective is sequence modeling.
Sources (4)
metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length

Profile review details

Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied.

Stable record: catalog-model-metagene-1

Specifications

Inputs, training, access and other details

Explanatory profile: limited source coverage · Automated source review, 2026-09-16. Review applies to the cited claims; unresolved fields are listed below. Numerical results retain their own review status.

Inputs, outputs and configuration
Property	Description and evidence
Model type	Autoregressive metagenomic transformer Sources (4) metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length
Architecture	Llama-style autoregressive transformer with 32 layers, width 4,096, 32 attention heads and a 1,024-token vocabulary in the inspected configuration. Sources (4) metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length
Inputs	DNA or RNA nucleotide sequences. Sources (4) metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length
Outputs	Sequence generation and representations for downstream metagenomic analyses. Sources (4) metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length
Parameters	7 billion. Sources (4) metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length
Known versions	METAGENE-1; exact model-card and configuration revision pinned in sources. Sources (4) metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length
Training data	More than 1.5 trillion base pairs sequenced from human wastewater samples, according to the model card. Sources (4) metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length
Training cutoff	The official pretraining README states that the wastewater corpus is not yet publicly released; a latest sample-collection date is not provided there. · Not reported in inspected sources Sources (4) metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length
Context limits	The released pretraining YAML sets max_seq_length=512. The model config lists max_position_embeddings=512 and max_sequence_length=2,048; these differing configuration fields do not establish a single validated inference maximum. Sources (4) metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length
Weights licence	Apache-2.0 declared in the model card. Sources (4) metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length
Access	Official downloadable model/card and usage examples: https://huggingface.co/metagene-ai/METAGENE-1 Sources (4) metagene-ai/METAGENE-1: README.md; metagene-ai/METAGENE-1: config.json; metagene-ai/metagene-pretrain: README.md; metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length
Code licence	Apache-2.0 Sources metagene-ai/metagene-pretrain: train/LICENSE · train/LICENSE: licence text

Evidence

Source checking verifies the cited claim or transcription. It does not establish independent reproduction.

Evidence table

Inspect claims, sources and review details

Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.

One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.

77 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-09-29-06401fd5b220
Property and statement	Original source and location	Review and provenance
Diagram caption Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation. Individual claims	metagene-ai/METAGENE-1: README.md Original source ↗ README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: ad8a1e0ee62b85058bfc05d823d8e8d4759edc48 Retrieved: 2026-09-16T19:46:20.736359+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `277517aa69527fccaeb85c124f3ec9772c5508fadb5c6ad423bee39a3071ad17` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram caption Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation. Individual claims	metagene-ai/metagene-pretrain: README.md Original source ↗ README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 82b9e142db2c7e0a268346d53344d6c3bf223066 Retrieved: 2026-09-16T20:04:02.447523+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `f09b3bc7f21c19d31ad90205da47010dc73b75096c12af5b350a925b4cac5a7b` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram caption Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation. Individual claims	metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml Original source ↗ README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 82b9e142db2c7e0a268346d53344d6c3bf223066 Retrieved: 2026-09-16T20:04:02.447523+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `0e764482ebf7e68c4751adfcf6430acdaa67fb683f59a31b9e25f8412d8ff5ad` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram caption Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation. Individual claims	metagene-ai/METAGENE-1: config.json Original source ↗ README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: ad8a1e0ee62b85058bfc05d823d8e8d4759edc48 Retrieved: 2026-09-16T19:46:20.736359+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `dc3751f8648b4ab24a3c0c42026267e1a2561b93249f54b460c0374e62d33f98` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram steps Nucleotide sequence BPE tokens Autoregressive transformer Sequence or representations Individual claims	metagene-ai/METAGENE-1: README.md Original source ↗ README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: ad8a1e0ee62b85058bfc05d823d8e8d4759edc48 Retrieved: 2026-09-16T19:46:20.736359+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `277517aa69527fccaeb85c124f3ec9772c5508fadb5c6ad423bee39a3071ad17` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram steps Nucleotide sequence BPE tokens Autoregressive transformer Sequence or representations Individual claims	metagene-ai/metagene-pretrain: README.md Original source ↗ README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 82b9e142db2c7e0a268346d53344d6c3bf223066 Retrieved: 2026-09-16T20:04:02.447523+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `f09b3bc7f21c19d31ad90205da47010dc73b75096c12af5b350a925b4cac5a7b` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram steps Nucleotide sequence BPE tokens Autoregressive transformer Sequence or representations Individual claims	metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml Original source ↗ README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 82b9e142db2c7e0a268346d53344d6c3bf223066 Retrieved: 2026-09-16T20:04:02.447523+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `0e764482ebf7e68c4751adfcf6430acdaa67fb683f59a31b9e25f8412d8ff5ad` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram steps Nucleotide sequence BPE tokens Autoregressive transformer Sequence or representations Individual claims	metagene-ai/METAGENE-1: config.json Original source ↗ README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: ad8a1e0ee62b85058bfc05d823d8e8d4759edc48 Retrieved: 2026-09-16T19:46:20.736359+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `dc3751f8648b4ab24a3c0c42026267e1a2561b93249f54b460c0374e62d33f98` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram title METAGENE-1 workflow Individual claims	metagene-ai/METAGENE-1: README.md Original source ↗ README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: ad8a1e0ee62b85058bfc05d823d8e8d4759edc48 Retrieved: 2026-09-16T19:46:20.736359+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.title` Source artifact SHA-256: `277517aa69527fccaeb85c124f3ec9772c5508fadb5c6ad423bee39a3071ad17` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram title METAGENE-1 workflow Individual claims	metagene-ai/metagene-pretrain: README.md Original source ↗ README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 82b9e142db2c7e0a268346d53344d6c3bf223066 Retrieved: 2026-09-16T20:04:02.447523+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.title` Source artifact SHA-256: `f09b3bc7f21c19d31ad90205da47010dc73b75096c12af5b350a925b4cac5a7b` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact

All evidence (CSV)All evidence (JSONL)How to read the evidence

Sources and history

View linked audit checks and correction history

Release 2026-09-29-06401fd5b220 · Record review: discovered

6 source records and release history

METAGENE-1 official resource · Original source · Not reported
metagene-ai/METAGENE-1: README.md · Original source · ad8a1e0ee62b85058bfc05d823d8e8d4759edc48
metagene-ai/METAGENE-1: config.json · Original source · ad8a1e0ee62b85058bfc05d823d8e8d4759edc48
metagene-ai/metagene-pretrain: README.md · Original source · 82b9e142db2c7e0a268346d53344d6c3bf223066
metagene-ai/metagene-pretrain: train/config_hub/pretrain/genomicsllama.yml · Original source · 82b9e142db2c7e0a268346d53344d6c3bf223066
metagene-ai/metagene-pretrain: train/LICENSE · Original source · 82b9e142db2c7e0a268346d53344d6c3bf223066

Download this release

Technical metadata and extraction receipts

Stable ID: catalog-model-metagene-1

areas: microbes-communities
method types: foundation model
entity level: family
version: 6B
reported name: METAGENE-1
access: Public Apache 2.0 checkpoint; 512-token context and large local memory requirement.
method type: foundation model
historical missing metadata: checkpoint revision: not_yet_extracted; training data: not_yet_extracted; licence: not_yet_extracted
metadata review scope: historical_missing_metadata preserves the original discovery state. Current descriptive evidence and missingness are recorded in profile.facts; numerical-result review is separate.
entity classification: review date: 2026-09-17; rationale: The cited profile describes a named learned biological predictor or representation model/family. Preserve this identity separately from task-specific fitting, individual checkpoints, pipelines and hosted access.; source ids: evidence-official-8e894fc6d180746a1808; evidence-official-f71e6c488025a4d82997; evidence-official-a6b46962aae5cfc348c6; evidence-official-b554679ea1503ce3d9f6; source locator: README.md: Model Overview and Usage; config.json; config.json: model_type, hidden_size, num_hidden_layers, num_attention_heads, vocab_size and sequence-length fields; official pretraining train/config_hub/pretrain/genomicsllama.yml: train.max_seq_length; ambiguities: None recorded

Related records

uses model: METAGENE-1

Suggest a correction