French AI startup Mistral AI has announced a new partnership with Agence France-Presse (AFP), marking another major collaboration between AI companies and traditional news media.
Under the agreement, Mistral's AI chatbot Le Chat will gain access to AFP's extensive news feed, which publishes around 2,300 articles daily across six languages - French, English, Spanish, Portuguese, German, and Arabic.
The integration aims to enhance Le Chat's responses with current, fact-based information. Mistral plans to roll out AFP content access to all users in the coming weeks.
The company is keeping the deal terms under wraps, but this type of partnership is becoming increasingly common. OpenAI has already secured nearly 20 similar media deals, and Meta recently added Reuters to their roster.
Other players have taken different approaches - Perplexity offers publishers a cut of advertising revenue, while Google simply uses news content in their AI summaries without compensation.
The great convergence
The AI chatbot landscape is starting to look remarkably uniform. All the major players - ChatGPT, Copilot, Le Chat, Perplexity, Meta AI, and Google Gemini - now offer essentially the same feature set: text generation, image creation, internet search, and news integration.
The one notable exception is Anthropic's Claude, which remains focused solely on text generation without internet access - though it maintains a slight edge in model quality.
Since GPT-4's debut, we haven't seen any revolutionary breakthroughs in large language models - just incremental improvements. This technological plateau has sparked an inevitable price war for API access. Making matters more competitive, open-source models are now flooding the market, offering even cheaper alternatives to the major AI labs' APIs.
Some AI companies appear to be betting on "reasoning models" and "agentic AI" as their next competitive moat. OpenAI is already testing this strategy by including their new o1 model in ChatGPT Pro subscriptions - though CEO Sam Altman admits they're still losing money on these premium offerings.
Once users become dependent on AI in their daily lives and locked into specific platforms, companies will have more leverage to raise prices. Microsoft is already testing these waters with an aggressive pricing model where agent-based AI can cost several hundred dollars per agent per day. OpenAI might follow suit - Altman has hinted they could switch to a usage-based pricing for ChatGPT in the future.