AI Model Language Support Checker

Different AI providers publish language support in different ways, and the differences matter before you commit time or money to a project. This tool encodes, for each flagship model, which languages the provider documents — and at what tier — so you can tell a formal support claim from an anecdote. It is a lookup, not a calculation: nothing is sent to any provider, and the same input always returns the same answer.

Every populated cell is drawn from a provider's own enumerated list and carries a link back to it. Silence in the docs is reported as "Not listed", never as an inferred "probably supports". Four tiers do all the work:

• Officially listed— the provider names the language in its supported-languages documentation. Meta's Llama model cards (8 languages for Llama 3.1, 12 for Llama 4), Google's 38-language Gemini text core, and Meta's NLLB-200 translation roster are all enumerated lists of this kind.
• Benchmarked only— the language appears in the provider's published evaluation set but not a formal support list. OpenAI's MMMLU (MMLU professionally translated into 14 languages) and Anthropic's 14-language multilingual table are benchmark sets: evidence the model was measured, not a promise it is supported.
• Speech: yes — the model documents transcription or synthesis for the language. OpenAI Whisper lists ~99 transcription languages; Gemini text-to-speech lists a broad set. Both include Sinhala and Tamil.
• Not listed — the provider does not document the language for that model. It may still work; the provider simply makes no claim, so neither do we.

The by-languageview answers "who supports this language?" across every model, filtered by task. The by-model view lists every language one provider documents for a model, with a count and the source link. A persistent Sinhala & Tamilcard surfaces the best-documented option per task for Sri Lanka's two national languages — the local question global tools sites ignore. A built-in integrity check confirms all 101 language codes resolve against the source table on every build, with no dangling references.

• OpenAI — Whisper LANGUAGES map (99 transcription languages, incl. si & ta)
• OpenAI — simple-evals (MMMLU / MGSM multilingual benchmark set)
• Anthropic — Claude multilingual support (14 benchmarked languages)
• Google — Gemini core language support (38-language text set)
• Google — Gemini text-to-speech supported languages (incl. si & ta)
• Meta — Llama 3.1 & Llama 4 model cards (supported languages)
• Meta — NLLB-200 / FLORES-200 language list (translation, incl. sin_Sinh & tam_Taml)

The language lists were last cross-checked against these official sources on 2026-07-12. The table is reviewed whenever a provider ships a new model family or updates a supported-languages list. This tool reports documented support only — it does not rate output quality or rank models; for that, see the related tools below.