Skip to content
Model infoModelsSnapshot: 15 Oct 2026

WRITER / Palmyra

InstructPalmyra 30B

Historical WRITER text model recorded in the 2023 HELM catalog. Reported parameters, release date and context are preserved with their evaluator provenance.

Language models (LLMs)
historicalClosed weightsText generation
Official documentation (opens in a new tab)Reviewed 14 Oct 2026
9 of 20 specification fields collected
Context window2K2,048 tokens
Maximum outputNot collectedAwaiting source review
Input billingNot collectedWRITER enterprise API (historical)
Output billingNot collectedNo inferred price

Specifications

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit
2,048 tokensSource for Context / input limit (opens in a new tab)

The historical evaluator describes this advertised context, then uses a safety margin of 2,000 tokens. The companion window-service factory explicitly maps this model to that service. This is not a separately verified current API limit.

Maximum input tokens
Not collected

Not verified for this exact historical release. No sibling or current-model specification inherited.

Maximum output
Not collected

Not verified for this exact historical release. No sibling or current-model specification inherited.

Reasoning
Not collected

Not verified for this exact historical release. No sibling or current-model specification inherited.

Tool calling
Not collected

Not verified for this exact historical release. No sibling or current-model specification inherited.

Structured output
Not collected

Not verified for this exact historical release. No sibling or current-model specification inherited.

Streaming
Not collected

Not verified for this exact historical release. No sibling or current-model specification inherited.

Batch processing
Not collected

Not verified for this exact historical release. No sibling or current-model specification inherited.

Fine-tuning
Not collected

Not verified for this exact historical release. No sibling or current-model specification inherited.

Context caching
Not collected

Not verified for this exact historical release. No sibling or current-model specification inherited.

Knowledge cutoff
Not collected

Not verified for this exact historical release. No sibling or current-model specification inherited.

Architecture
Human-feedback reinforcement-learning instruction tuningSource for Architecture (opens in a new tab)

Training description from HELM; underlying architecture details not disclosed here.

Languages
Not collected

Not verified for this exact historical release. No sibling or current-model specification inherited.

Parameters
30,000,000,000Source for Parameters (opens in a new tab)

Reported by the original HELM evaluator metadata.

Release date
2023-02-16Source for Release date (opens in a new tab)

Evaluator-reported release date; not an independently verified provider launch date.

Input modalities
TextSource for Input modalities (opens in a new tab)
Output modalities
TextSource for Output modalities (opens in a new tab)
Release identity
HELM identity writer/palmyra-instruct-30Source for Release identity (opens in a new tab)

Original research identity; not asserted to be an alias of a later commercial release.

Generation controls
prompt, maxTokens, temperature, topP, bestOf and stopSource for Generation controls (opens in a new tab)

Historical evaluator request fields. No supported ranges or defaults are established; multiple completions were implemented with repeated requests.

API request fields
Organization-scoped enterprise completions request with the model engine in the route.Source for API request fields (opens in a new tab)

The evaluator used its own organization identifier. That identifier is not a public shared endpoint or customer configuration.

Access & pricing

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

WRITER enterprise API (historical)

Provider source (opens in a new tab)
API model ID
palmyra-instruct-30

Historical evaluator forwards the final model-engine name in the organization-scoped completions route.

Access
Original enterprise API credentials and organization access required. Current access and final shutdown date unverified.
Regions
Not collected

Endpoint pricing has not been collected. Check the provider for applicable rates.

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

Caching

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

WRITER enterprise API (historical)

palmyra-instruct-30
Not verified

No exact-model provider cache contract verified. Evaluator request caching and comments about random_seed do not establish prompt caching, TTL or token discounts.

How it works
Not collected

No exact-model provider cache contract verified. Evaluator request caching and comments about random_seed do not establish prompt caching, TTL or token discounts.

Minimum reusable input
Not collected

No exact-model provider cache contract verified. Evaluator request caching and comments about random_seed do not establish prompt caching, TTL or token discounts.

Lifetime & refresh
Not collected

No exact-model provider cache contract verified. Evaluator request caching and comments about random_seed do not establish prompt caching, TTL or token discounts.

How to enable it
Not collected

No exact-model provider cache contract verified. Evaluator request caching and comments about random_seed do not establish prompt caching, TTL or token discounts.

Usage fields
Not collected

No exact-model provider cache contract verified. Evaluator request caching and comments about random_seed do not establish prompt caching, TTL or token discounts.

Billing rules
Not collected

No exact-model provider cache contract verified. Evaluator request caching and comments about random_seed do not establish prompt caching, TTL or token discounts.

Limits & data boundaries
Not collected

No exact-model provider cache contract verified. Evaluator request caching and comments about random_seed do not establish prompt caching, TTL or token discounts.

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Caching reviewed 14 Oct 2026

What the measurements say

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

The record behind the model

Why it is included

Closed-weight distribution inference: the original evaluator records limited access through WRITER enterprise inference, and the official public artifact inventory has no matching exact trained checkpoint. Public small Palmyra models and InstructPalmyra 20B are distinct artifacts, not evidence that these original weights were released.

Classification evidence (opens in a new tab)

Lifecycle dates

Release date reported by HELM16 Feb 2023Source (opens in a new tab)

Sources reviewed