--- title: "Deepgram's cost floor and Mistral's price floor landed within 24 hours" description: "AWS's engineers built Deepgram's cost floor, then open-sourced the tool that gives its rivals the same floor, while Mistral published a price list that undercuts Deepgram. Amazon wins either way; Deepgram's investors wait." publisher: "The Inference" section: "Model" published: 2026-10-24T15:17:01.501Z modified: 2026-10-24T15:17:01.501Z canonical: https://theinference.org/article/deepgram-s-cost-floor-and-mistral-s-price-floor-landed-within-24-hours language: en keywords: "Voice AI, Mistral AI, Deepgram, AWS, Pricing, Chips" --- # Deepgram's cost floor and Mistral's price floor landed within 24 hours > AWS's engineers built Deepgram's cost floor, then open-sourced the tool that gives its rivals the same floor, while Mistral published a price list that undercuts Deepgram. Amazon wins either way; Deepgram's investors wait. Deepgram transcribes speech for more than 200,000 developers and 1,400 organizations, processing quadrillions of audio tokens a year [1]. This week, two documents reset what a minute of that speech costs from both directions, and the two have not been read together until now. ## The cushion Amazon just built On October 19, [AWS published a case study](https://aws.amazon.com/blogs/machine-learning/how-deepgram-optimized-voice-ai-at-scale-with-aws-trainium/) of Deepgram's Nova-3 speech-to-text model ported onto Trainium2, [Amazon](https://theinference.org/markets/companies/amazon)'s in-house AI chip [1]. Eight model replicas run on a single trn2.3xlarge instance, and on Deepgram's own offline batch workloads the setup processed 11 to 70 percent more audio per dollar than the compute Deepgram had been running [1]. Deepgram delivered Nova-3 as standard PyTorch and ported it over several months together with AWS's Forward Deployed Engineering team [1]. One day later, at the PyTorch Conference, AWS open-sourced TorchNeuron, the tool that made the port possible, now a one-line change for any PyTorch app, with more than 30 early customers including Deepgram's voice rival Cartesia [3]. Colin Brace, VP of Technology Software at AWS Annapurna Labs, the unit that makes Amazon's custom chips, [told TechCrunch](https://techcrunch.com/2026/10/19/amazon-is-taking-its-biggest-shot-at-nvidia-yet-with-new-torchneuron-dev-tool/) to look at Deepgram, "processing 70% more audio per dollar" [2]. The reference customer is also the sales pitch. Note the range is company-reported, never independently reproduced [2]. ## The price floor arrives from Paris On October 20, [Mistral launched Voxtral 3](https://mistral.ai/news/voxtral-3/) and named Deepgram as a vendor it undercuts: batch transcription at $0.0035 a minute on a 4-billion-parameter model, speech generation at $13 per million characters, a full voice agent API at $0.04 a minute [4]. That is eleven times the transcript price; the margin in voice AI is migrating up the stack [5]. *Chart: **The voice agent API costs eleven times what the transcript costs** Bar chart of Mistral Voxtral 3 per-minute prices: batch transcription $0.0035, flagship transcription $0.0040, realtime transcription $0.008, voice agent with Mistral Small 4 $0.04, voice agent with larger model $0.08* *Published per-minute list prices from Mistral's Voxtral 3 launch, October 20, 2026, US dollars. Text-to-speech is priced separately at $13 per million characters and is not shown. Source: Mistral [4]. [4]* ## Whose money moves Deepgram is private, so the exposure runs through Amazon and the chip race. TorchNeuron attacks [Nvidia](https://theinference.org/markets/companies/nvidia)'s CUDA lock-in, where AWS says roughly 85% of AI apps live [2]. Amazon wins either way: the case study sells Trn2 capacity, and the tool makes that capacity a one-liner for Deepgram's competitors too [5]. The cost floor Deepgram earned is now a cushion against a price cut [Mistral](https://theinference.org/markets/companies/mistral-ai) has already made, not a moat. Watch the first independent benchmark of Voxtral's accuracy claims; [the price list is real](https://shattered.io/mistral-voxtral-3-speech-ai-pricing-2026/), the quality comparisons are Mistral's own [5]. ## Takeaway AWS's Deepgram case study gave the voice-AI vendor a cost edge on Trainium, then AWS open-sourced TorchNeuron to every rival, and Mistral's Voxtral 3 price list undercut Deepgram's transcript price: the cost floor is a cushion, not a moat. ## Sources 1. [AWS Machine Learning Blog, 'How Deepgram optimized voice AI at scale with AWS Trainium', 19 October 2026](https://aws.amazon.com/blogs/machine-learning/how-deepgram-optimized-voice-ai-at-scale-with-aws-trainium/) 2. [TechCrunch, 'Amazon is taking its biggest shot at Nvidia yet with new TorchNeuron dev tool', 19 October 2026](https://techcrunch.com/2026/10/19/amazon-is-taking-its-biggest-shot-at-nvidia-yet-with-new-torchneuron-dev-tool/) 3. [CryptoBriefing, 'Amazon gives Trainium3 native PyTorch support to win over AI researchers', 23 October 2026](https://cryptobriefing.com/amazon-trainium3-native-pytorch-torchneuron/) 4. [Mistral AI, 'Voxtral 3', 20 October 2026](https://mistral.ai/news/voxtral-3/) 5. [shattered.io, 'Mistral's Voxtral 3 Debuts at $0.0035 a Minute', 22 October 2026](https://shattered.io/mistral-voxtral-3-speech-ai-pricing-2026/) --- Deepgram's cost floor and Mistral's price floor landed within 24 hours — The Inference. Canonical: https://theinference.org/article/deepgram-s-cost-floor-and-mistral-s-price-floor-landed-within-24-hours