Try Kling Image V4 Long for Page Mockups from Kling AI →
Models
Agents
Workflows
Studio
PricingBlogDocs
ExploreDiscover models by categoryDiscoverModels and effects built by WiroBrowse All ModelsBrowse the complete catalogSee FavoritesSign in to view saved models
Generative Media AgentCreate and edit media by chattingWorkflow AgentBuild visual workflows with Agent
OverviewThe platform at a glanceLearnSkills, knowledge, guardrailsAnatomyWhat makes agents reasonBuild Your AgentPick connectors and skills, choose a plan, deploy
Pre-built AgentsBrowse the catalog
Agent Usecases
Ad Campaign ManagerApp Event ManagerApp Review RepliesBarber BookingCustomer Win-BackEcommerce ListingsRestaurant Reviews
Sign InStart Building
Explore models/
LLM & ChatActive

EuroLLM 1.7B Instruct

byutter-project

The first instruction-tuned model in the EU-funded EuroLLM series: 1.7B parameters, tuned on EuroBlocks for translation and general instructions in 35 languages, including all 24 official EU languages.

ChatLLMRAGEurollm ChatBf16
Model ID
EuroLLM-1.7B-Instruct
Provider
utter-project
Updated
Oct 29, 2026, 20:32
EuroLLM-1.7B-Instruct
0
Comments
Average rating : 0 (0 users)
Providerutter-project
ModelEuroLLM-1.7B-Instruct
ChatLLMRAGEurollm Chat
wiro playground—utter-project/EuroLLM-1.7B-Instruct

Prompt to send to the model.

Your request will cost $0.0006 per second.

Running this model on Wiro costs approximately $0.0236 in total.

(Total cost varies depending on the request's execution time.)
Sample outputs
Loading preview…
utter-project-eurollm-1-7b-instruct-sample-1.txt
Updated Oct 29, 2026, 20:32

Overview

EuroLLM-1.7B-Instruct is the first instruction-tuned release of EuroLLM, an EU-funded project building open language models for every official EU language. Nine organisations built it, among them Unbabel, Instituto Superior Técnico and the University of Edinburgh.

The base model, EuroLLM-1.7B, was pre-trained on 4 trillion tokens of web text, high-quality datasets and parallel text to and from English. Instruction tuning used EuroBlocks, which centres on general instructions and machine translation. It is a dense 24-layer transformer with 1.657B parameters.

The 35 languages are Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hungarian, Irish, Italian, Latvian, Lithuanian, Maltese, Polish, Portuguese, Romanian, Slovak, Slovenian, Spanish and Swedish, plus Arabic, Catalan, Chinese, Galician, Hindi, Japanese, Korean, Norwegian, Russian, Turkish and Ukrainian.

The EuroLLM team reports a FLORES-200 translation average of 86.89 for this model, against 81.59 for Gemma-2B and 85.27 for Gemma-7B, both tuned on the same EuroBlocks data.

What you can build

  • A first translation pass for UI strings, support macros or short documents, followed by human review.
  • Small assistants that rewrite or simplify text in Polish, Greek or Swedish.
  • The answer step of a retrieval pipeline: pasted passages carry the facts, the model replies in the user's language.

Inputs

Only the prompt is required. An optional system prompt goes in front of it; the default asks for helpful, honest and safe answers.

Send the same user_id and session_id with each request to keep chat history; a new session_id starts a fresh one.

Sampling and quantization are both on by default. Temperature starts at 0.7 (0 to 5), top_p at 0.95, top_k at 0 and the repetition penalty at 1.0. Stop sequences take a semicolon-separated list, and the seed defaults to 123456.

Outputs

A plain text reply. The sample on this page answers the default Tower of London prompt with five numbered battles, each followed by the same story hook. Hastings appears twice, and none of the five battles was fought near the Tower of London.

Recommended settings

  • For translation, turn do_sample off and follow the EuroLLM team's prompt layout: an instruction naming both languages, the source after a label like "English:", then an empty "Portuguese:" label.
  • With max_tokens at 0, the default, the reply budget is about 2,048 tokens, half of the 4,096-token window. The team's own example translates one sentence with a 20-token limit, and a lower max_tokens leaves more of the window for input.
  • When list items repeat, as in the sample, raise the repetition penalty slightly above 1.0.

Limitations

  • The card gives a 4,096-token sequence length, shared by prompt, history and reply. On Wiro part of that window is held for the reply, so the input limit is lower than 4,096 tokens. Over it, the oldest turns are dropped first, then the end of the newest message is cut, so keep documents and retrieved passages short.
  • For the pre-trained model, the team reports HellaSwag results close to Gemma-2B and weaker ARC Challenge scores, which it links to the smaller number of non-embedding parameters.
  • Expect factual slips on open questions, as in the sample. Paste source text when accuracy matters.

Safety & compliance

Released under the Apache License 2.0. The card states the model has not been aligned to human preferences and may produce hallucinations, false statements or harmful content. The default system prompt asks for safe answers but filters nothing, so moderate replies before they reach end users.

API quick start

Run EuroLLM-1.7B-Instruct with a single API call.

POST https://api.wiro.ai/v1/Run/utter-project/eurollm-1-7b-instruct
{
  "prompt": "What are some interesting historical even…",
  "user_id": "...",
  "session_id": "...",
  "system_prompt": "You are a helpful, respectful and honest …"
}
curl
curl -X POST "https://api.wiro.ai/v1/Run/utter-project/eurollm-1-7b-instruct" \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_WIRO_API_KEY" \
  --data-binary @- <<'JSON'
{
  "prompt": "What are some interesting historical even…",
  "user_id": "...",
  "session_id": "...",
  "system_prompt": "You are a helpful, respectful and honest …"
}
JSON

Other models you might like

Explore similar llm & chat models.

View all
EuroLLM-9B-Instruct
Chat

EuroLLM-9B-Instruct

utter-project

aya-expanse-8b
Chat

aya-expanse-8b

cohereforai

Mistral-Nemo-Instruct-2407
Chat

Mistral-Nemo-Instruct-2407

mistralai

CodeLlama-7b-Instruct-hf
Chat

CodeLlama-7b-Instruct-hf

meta-llama

  • Llama-3.2-3B-Instructmeta-llama
  • Qwen2.5-7B-Instructqwen
  • Phi-3.5-mini-instructmicrosoft
  • Qwen3-Coder-30B-A3B-Instructqwen
  • Mathstral-7B-v0.1mistralai
  • rag-chat-websitewiro
  • Qwen2.5-3B-Instructqwen
  • deepseek-math-7b-instructdeepseek-ai

Discover, test, and run AI models, build workflows and agents with one unified API.

All systems operational
WiroModelsAgentsMCP ServerLLM GatewayBrowser ExtensionIntegrationsPricingAboutBlogCareersContactPartnerChangelogStatusFAQ
CategoriesText to ImageImage to ImageImage EditingImage to TextText to VideoImage to VideoVideo to VideoText to SpeechSpeech to TextSpeech to SpeechVoice CloneRealtime ConversationMusic GenerationLLMUtilities
ModelsNano Banana 2GPT Image 2.5Seedream V5 ProSeedance 2.5Veo 3.1Kling V3FLUX 3FLUX.2 ProWan 3.0 PrimeGrok Imagine 1.5
PartnersGoogleOpenAIByteDanceBlack Forest LabsKling AIQwenAlibabaxAIMiniMaxElevenLabs
Getting StartedWiro ModelsWiro EffectsIntroductionAuthenticationProjectsCode ExamplesWiro MCP ServerSelf-Hosted MCPn8n IntegrationLLMs.txt
API ReferenceModelsRun a ModelModel ParametersTasksLLM & Chat StreamingWebSocketRealtime VoiceFiles
© 2026 Wiro AI. All rights reserved.
PrivacyTermsData Deletion