Some modes treats input file text as prompt instructions, causing unintended translations #503

New Issue

2026-02-16T00:18:03-05:00

yindo commented

2026-02-16 00:18:03 -05:00

Originally created by @adreichert on GitHub (Jul 14, 2025).

Summary

Some LlamaParse modes inadvertently treat text from input files as part of the prompt, as illustrated in the example below. Ideally, LlamaParse itself should include explicit guidance in its default system message, clarifying how input file text should be handled.

Reproduction

Go to https://cloud.llamaindex.ai/
Parse paris.pdf in premium. The file is in English but the output text is in French

Originally created by @adreichert on GitHub (Jul 14, 2025). # Summary Some LlamaParse modes inadvertently treat text from input files as part of the prompt, as illustrated in the example below. Ideally, LlamaParse itself should include explicit guidance in its default system message, clarifying how input file text should be handled. # Reproduction * Go to https://cloud.llamaindex.ai/ * Parse [paris.pdf](https://github.com/user-attachments/files/21222039/paris.pdf) in premium. The file is in English but the output text is in French

yindo added the bug label 2026-02-16 00:18:03 -05:00

Sign in to join this conversation.

Branches Tags

main

claude/slack-session-O3953

dependabot/github_actions/slackapi/slack-github-action-3.0.3

dependabot/github_actions/pnpm/action-setup-6

cursor/llamaparse-heartbeat-timeout-retry-d4fc

cursor/s3-transaction-race-6a10

cursor/llamacloud-chart-version-0-7-1-2dc9

nprad/extract-tests

changeset-release/main

nprad/update

nprad/xlsx

logan/deprecate_llama_parse

nprad/e2e-test-freq

nprad/fix-e2e-1

nprad/fix-e2e

nprad/e2e-test-invalidate-cache

nprad/slack-msg

nprad/retries

neerajprad-patch-3

nprad/reparse-yml2

nprad/reparse-yml

nprad/hourly-extract-tests

nprad/bump-patch

nprad/bump-version

claude/refactor-invalid-extraction-error-j6fJs

pierre/supportNoExtensionFiles

pierre/allowNotExtFiles

adrian/forgot-ts-changeset

adrian/bbox-support

logan/deprecate_notice

logan/fix_publish

clelia/corrent-links-in-readme

logan/improve_polling

parse-example

logan/client_tier

javier/li-4421-notebook-demo-using-split-api

fix-sheets-api

add_sheets_options

nprad/bump-v

logan/line_level_bbox

logan/dummy_release

logan/no_presigned_url

timeout

roman/batch-parse-example

adrian/testing-utils

clelia/agent-data-mock

pierre/moreParseParameters

cursor/implement-test-plan-and-create-mocked-tests-gpt-5.1-codex-high-3eb6

dependabot/github_actions/actions/checkout-6

cursor/implement-llama-cloud-testing-utilities-and-tests-gpt-5.1-codex-high-ae1a

nprad/nb-extract

logan/propagate_retrieval_mode

logan/beta_spreadsheet

adrian/fix-npm-build

adrian/ts-params

adrian/no-org-classify

nprad/bump-0.6.78

pjames27/job-id-on-parse-failure

nprad/sourcetxt

clelia/add-classify-ts

dependabot/github_actions/actions/setup-node-6

jerry/fix_classify_nb

bogdan/add-aggressive-table-extraction

logan/safest_types_possible

logna/fix_default_bbox_values

pjames27/only-escape-single-dollar

adrian/re-enable-js

logan/loosen_packaing

terry/fix_citataion_type

adrian/fixup-tagging

logan/fix_changeset_harder

logan/fix-version

adrian/fix-tag

pjames27/escaping-notebook-md-dollarsigns

adrian/tmp-disable-ts-release

adrian/changesets-fixes

adrian/fix-token

adrian/nvmrc

adrian/changesets

adrian/update-llama-cloud-dep

mnl/parse-classify-extract

cursor/LI-3624-handle-validation-errors-for-agent-data-retrieval-7504

adrian/extraction-metadata-bug

cursor/LI-3623-add-failing-test-for-citation-field-bug-9352

adrian/no-org-id

cursor/LI-3577-refactor-agent-fields-in-llama-cloud-services-1796

mnl/getting_started_multimodal

parse-error-handling

jerry/add_classify_extract_nb

mnl/add-streaming-citations

pierre/ChartAndBbox

nprad/v0.6.64

nprad/remove-report

clelia/fix-ts-release-notes

adrian/fix-v-prefix

adrian/client-unification

adrian/lint-version

jerry/new_getting_started_notebook

surface-errors

jerry/fix_preset_nb

logan/update_notebooks

jerry/add_parse_preset_notebooks

adrian/override-base-url

logan/deprecate_some_methods

jerry/fix_composite_retriever

bump-llama-cloud-services-0-1-38

adrian/extract-file

adrian/handle-error-field

fix/export-extracted-field-metadata-types

adrian/xdist

adrian/fix-lost-dimension

logan/sec_fixes

logan/fix_audio_results

nprad/update-extract-doc

clelia/refactor-parse-ts

clelia/add-extract-to-ts

clelia/changesets

nprad/update-extract

clelia/fix-python-tests

llama-cloud-index-demo

ms/add-e2e-tests

nprad/claude

ms/remove-llamaindex

adrian/re-enable

adrian/citation-confidence-fixes

adrian/extracted-citations

clelia/restructure-docs

clelia/delete-symlink-examples

clelia/examples-symlink

clelia/add-changesets

clelia/ts-examples

logan-markewich-patch-1

clelia/no-git-checks-on-publish

clelia/add-ts-examples

clelia/ts-refactoring

clelia/add-llamacloud-index

clelia/fix-npm-release

adrian/citations

javier/bump-llamacloud-version

clelia/monorepo

multimodal-reportgen-agent

adrian/derp

adrian/public-agent-data

update_get_result

adrian/update-0.6.49

adrian/full-md

nprad/update-v0.6.48

clelia/v0.6.47

clelia/add-table-extraction

rebelde/deprecate-llama-parse

adrian/agent-file-fields

pierre/merge_tables_across_pages_in_markdown

adrian/agent-data

logan/relax_pydantic_job_object

nprad/v0.6.42

pierre/headersFooters

nprad/bump-v0.6.39

logan/except_more

add-rpe

nprad/bump-v0.6.37

nprad/fix-flaky

nprad/v0.6.36

nprad/retry-job-fetch

nprad/v-0.6.35

logan/fix_partition

pierre/high_res_ocr

pierre/addWarningOnUnusedParameter

logan/bump_llama_cloud

jerry/add_fund_analysis_nb

nprad/bump-v0.6.30

suo/allow_none_project

logan/more_optional_types

logan/v0.6.25

pierre/outlined_table_extraction

logan/v0.6.24

nprad/insider-trans

nprad/bump-v-0.6.23

fix-verify

logan/v0.6.22

extract-with-citations

logan/more_optional

logan/v0.6.20

sacha/v0.6.19

pierre/auto_mode_configuration_json

pierre/preset

logan/original_width_height_optional

bump-v0.6.16

nprad/upgrade-le

logan/nits

pierre/page_error

logan/llama-report-tests

sacha/feat/add-markdown_table_multiline_header_separator

nprad/file-col

logan/parse_readme_nits

logan/v0.6.12

logan/new_result_object

nprad/le-docs

bump-v0.6.11

logan/v0.6.10

nprad/text-input-support

pierre/add_compact_markdown_table

nprad/pooling

nprad/support-bytes

marplex/parse_layout_agent_demo

neerajprad-patch-2

jerry/add_dd_notebook

nprad/rename-test-endpoint

jerry/add_solar_panel

mnl/eu-saas-readme

jerry/fix

jerry/extract_demo

bump-v0.6.9

pierre/update_parse_mode

nprad/nb-update

bump-v0.6.7

nprad/httpx-params-only

nprad/custom-httpx-client

nprad/xfail-report

bump-v0.6.6

nprad/update-lc

george/parse-retry

ljv/sec-edits

nprad/sec-10k

bump-v0.6.5

sacha/feat/adaptive_long_table

bump-0.6.3

neerajprad-patch-1

logan/vbump

nprad/add-le

pierre/upParameters

logan/fix_workflow

logan/org_id

jerry/add_gemini2_flash_nb

logan/v0.6.0

logan/refactor_services

sacha/fix/release-pipeline

pierre/newParameters

logan/v0.5.19

pierre/newFeatures

jerry/update_auto_mode

ljv/auto-mode-extended

jerry/auto_mode_notebook

pierre/htmlHideNavigationElements

ljv/json-mode-tour

pierre/morearguments

logan/add_test

jerry/dynamic_section_retrieval

pierre/input_url

pierre/xlsx2

pierre/newParams

jerry/mm_report_gen_image

sacha/feat/continuous-mode

sacha/chore/gh-template-update

jerry/fix_rfp_example

jerry/add_annotate_links

sacha/feat/formatting-instruction

sacha/feat/add-missing-parameters

jerry/add_rfp_workflow

jerry/add_mm_contextual_retrieval

logan/check_error

jerry/fix_excel_nb

jerry/move_o1_excel

pierre/premium_mode

jerry/fix_mm_slide_deck

sacha/fix/result-type-no-json

pierre/fix_path

sour/bump

jerry/update_readme

suo/take_screenshot

jerry/add_report_gen_agent

jerry/add_report_example

jerry/small_edits

jerry/add_gpt4omini_example

pierre/fix_page_separator

jerry/nit_fix_multimodal

jerry/add_sonnet

jerry/nit_fix

jerry/add_mm_notebook

pierre/new_param

logan/v0.4.5

pierre/new_params

jerry/dcf_rag_v2

jerry/add_split_by_page

jerry/rewrite_advanced_example

pierre/new_options

jerry/add_dcf_excel

jerry/add_kg_agent

pierre/spreadsheet

fastMode

logan/add_features

fix_img_path

jerry/add_caltrain_weekend_doc

jerry/add_gpt4o_mm_rag

jerry/fix_badge

jerry/caltrain_schedule

jerry/gpt4o_nb

logan/new_params

logan/fix_image_extension

html

html_support

update-notebooks

logan/fix_tests

logan/v0.4.1

logan/qol_fixes

logan/0.4.0

pierre/extend_formats

pierre/wip

jerry/fix_mongodb_nb

jerry/fix_advanced_title

jerry/revert_revert

hz/json

logan/openai_agent

jerry/fix_grammar

new_demo

ljv/parsing-instructions-demo

jerry/add_financial_powerpoint

jerry/add_ppt_demo

jerry/support_more_file_types

logan/fix_json_bug

jerry/add_json_cookbook

pierre/json-images

logan/fix_language

jerry/add_table_comparisons

jerry/bump_version_0.3.5

jerry/add_language_cookbook

pierre/add_language_support

jerry/new_example_astra

jerry/fix_astra_example

logan/async_batch

hz/demo_v5

hz/demo_v4

logan/update_docs

jerry/fix_v10

hz/demo_V3

hz/demo_v1

hz/llama_p_demo

suo/bump_version

suo/handle_exception

hz/demo_v2

hz/hz_pdf

sour/print_job_id

jerry/nits

hz/demo

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: run-llama/llama_cloud_services#503