The Commons LabLanguage modelOpen weights
Hemmingway-1
Hemmingway-1 has no overall index value yet because there are too few independent benchmark results.
37out of 100new calculation
based on two data groups, mostly search, range 27 to 46
Summary
Hemmingway-1 focuses on everyday English writing such as messages, emails, and difficult conversations. The Commons Lab post-trained Qwen3.8-27B to deliver the requested text directly, without surrounding it with preambles and alternatives.
The model has 27 billion parameters and handles 262,144 context tokens. Its weights use CC BY-NC 4.0, with commercial use requiring a separate agreement according to its model card.
On this page
All benchmark results
Every published result with its source, test setup and comparison models.
Show all 6 published benchmark results
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
Hemmingway-1 Human-Likeness, hard-asks subgroup (Show measurement, test conditions, and source)
- Source value
- 72
- Metric
- judge preference share selecting the response as human-written
- Unit
- %
- Benchmark version
- Altworld Hemmingway-1 model card, Where it wins section
- Category
- Human-Likeness preference
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Altworld official Hemmingway-1 model card
- Status
- active
- Retrieved at
- 2026-10-18
- Methodology
- Vendor-reported result. Altworld reports a blind head-to-head Human-Likeness judgment on the hard-asks prompt subgroup. Higher means the judge more often selected the model's response as the one written by a person. Matchups were run in both orders and judged by a different model from those being compared. Hemmingway-1 is the 27B open-weight model built on Qwen3.8-27B. The judge identity and subgroup size are not stated.
- Context
- This is a developer-reported, internally built benchmark, not an independent evaluation. The card's 80 real requests refer to the broader message comparison and are not assigned as the hard-asks subgroup sampleCount. No subgroup denominator or confidence interval is published. The existing The Message, Not a Memo chart score is a separate metric.
CommunicationBench (Show measurement, test conditions, and source)
- Source value
- 1,026
- Metric
- CommunicationBench reported score
- Unit
- No unit provided
- Category
- source-specific
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Provider model card
- Status
- active
- Retrieved at
- 2026-10-04
- Methodology
- Vendor-reported result. Author-built internal benchmark; 80 real requests, blind head-to-head answers to the same request, shuffled presentation, both answer orders, and a judge model different from the judged models.
- Context
- Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.
Human-Likeness (Show measurement, test conditions, and source)
- Source value
- 1,032
- Metric
- Human-Likeness reported score
- Unit
- No unit provided
- Category
- source-specific
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Provider model card
- Status
- active
- Retrieved at
- 2026-10-04
- Methodology
- Vendor-reported result. The same author-run blind matchups as CommunicationBench, with the judge asked which answer a person wrote.
- Context
- Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.
EQ-Bench 4 (Show measurement, test conditions, and source)
- Source value
- 1,330
- Metric
- EQ-Bench 4 reported score
- Unit
- No unit provided
- Benchmark version
- EQ-Bench 4
- Category
- source-specific
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Provider model card
- Status
- active
- Retrieved at
- 2026-10-04
- Methodology
- Vendor-reported result. The card identifies EQ-Bench 4 as public and says it was run with its own harness; the value is the Hemmingway-1 bar read from the complete chart.
- Context
- Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.
StoryBench (Show measurement, test conditions, and source)
- Source value
- 1,197
- Metric
- StoryBench reported score
- Unit
- No unit provided
- Category
- source-specific
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Provider model card
- Status
- active
- Retrieved at
- 2026-10-04
- Methodology
- Vendor-reported result. Author-built internal creative-writing benchmark; chart shown in the primary model card, with the same disclosed blind-comparison approach.
- Context
- Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.
The Message, Not a Memo (Show measurement, test conditions, and source)
- Source value
- 39
- Metric
- raw chart score on 0-100 axis; denominator not stated
- Unit
- No unit provided
- Category
- writing
- Direction
- Lower is better
- Source type
- vendor-reported
- Evaluator
- Altworld official Hemmingway-1 model-card chart
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Vendor-reported result. Exact model Altworld/Hemmingway-1. The card defines the measured behavior as how often a model buries the requested text in commentary, options, and notes. The chart title is ‘The Message, Not a Memo’, its subtitle says ‘Replies wrapped in commentary or options. Lower is better’, and it marks the evaluation INTERNAL. It shows a 0-100 axis and a Hemmingway-1 value of 39, but no denominator, sample count, benchmark set, model-call settings, run date, or uncertainty.
- Context
- Altworld marks this chart INTERNAL. The card's 80 real requests are stated for the separate head-to-head message evaluation and are not used as this row's N. No exact denominator or uncertainty is published.
Popularity compared
How often the model is searched for, used and mentioned, compared with every model of its category.
Popularity index
The 100 most popular models of this category, 0 to 100, higher is more popular
20 of 101 models
As of November 2026 · Source: gradually.ai popularity index
Over time
Weekly values from October 18, 2026 to November 20, 2026
Hemmingway-1Claude Opus 4.6 37GLM-4.5V 37Kimi K2 37Nemotron Nano 2 9B 37
Method changes in the timeline, a jump at these points is no market event:
- 1November 1, 2026: Value only with search or usage data, 30-day attention
- 2November 1, 2026: Method version 4 (common scale, shrinkage)
- 3November 9, 2026: New topics measurement V20 (daily quorum, short forms)
What is gaining popularity right now, across models, vendors, harnesses and AI subscriptions, is collected in All AI trends.
Similar models
More models of the same kind as Hemmingway-1, chosen by model line, developer, size, price and release date.
All specifications
Architecture
Input and output
Watermarking and provenance
AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.
- AI watermarking
- Unknown
- no source
- Watermarking technique
- Not known
- no source
Licence and access
- Shutdown
- None announced
- no source
Frequently asked questions about Hemmingway-1
How large is the context window of Hemmingway-1?
262,144 tokens. Separate limits may apply to input and output.
Can I run Hemmingway-1 locally?
Yes, the weights are open.
Stated by The Commons Lab
- Open weights
- yes
- License
- CC BY-NC 4.0. Commercial use requires a separate agreement.
- Parameters
- 27B
- active per token
- n/a
Sources and data date
Show all 16 sources
Benchmark sources at a glance
Which independent evaluators tested this model. This is coverage, not a quality ranking.
No active result from the reviewed external evaluators can currently be attributed to this model. Other source evidence remains available below.
Without vendor claims and historical or disputed results; those stay labelled in the single values.