Midcentury

What is Midcentury?

Midcentury supplies data and simulation infrastructure for teams building robots and other physical AI. Its catalog includes first-person video of manual work in industrial and everyday settings, paired with IMU, audio, 3D hand pose, depth and task annotations, plus head and wrist recordings of robot-gripper tasks. The company also sells speech datasets, both ready-made and collected to order, in languages such as Hindi, Bengali, Japanese and Luxembourgish. Midcentury Matrix is its simulation product: you build digital twins from real deployment conditions, run large numbers of test scenarios in parallel on GPUs, replay failures and turn them into new training material. Typical customers are robotics, foundation-model and speech AI teams that need training data at scale. Pricing is not published. Cost and delivery time depend on language, volume, licensing terms and technical specifications, so you request a quote from the company.

Use cases

• Pretrain robot policies on human action data
• Evaluate robots in simulation before hardware
• Train speech recognition models
• Build digital twins of deployment sites
• Replay robot failures as new scenarios
• Commission custom datasets

Who it's for

• Robotics teams
• AI research labs
• Speech AI teams

Why use Midcentury

• Gets first-person manual-work video with hand pose and action labels from one supplier rather than recording it yourself.
• Lets you test robot policies across many simulated scenarios before running them on hardware.
• Covers speech data in several Indian and other languages, including code-mixed speech such as Hinglish.
• Custom collections are configured to your spec, so datasets match the domain you are training for.

Midcentury features

• First-person video with IMU, audio, 3D hand pose, point tracks and depth
• Task and action labels on egocentric footage
• Gripper-task recordings with RGB and stereo cameras, TCP pose and gripper state
• Midcentury Matrix simulation platform with digital twins
• Parallel GPU evaluation across thousands of scenarios with failure replay
• Ready-made conversational and customer-service speech datasets
• Speech collected to order by language, sample rate, channels and speaker profile
• Time-aligned transcripts, verbatim or normalized text, delivered in your format
• Custom dataset design on request

How to get started with Midcentury

1. Describe the capability you are training, such as manipulation, robot evaluation or speech recognition.
2. Pick a ready-made dataset from the catalog or ask for a custom collection with your specs.
3. Agree volume, licensing terms, annotation depth and quality checks with the Midcentury team.
4. Receive audio, video and annotations in your preferred format.
5. For robot testing, use Midcentury Matrix to design, test and train in simulation and feed failures back into your policies.

Midcentury limitations

• Pricing is quoted per project and is not published
• The robotics manipulation dataset is still being collected
• Hindi transcripts come from an in-house speech model, with human review scoped separately

Midcentury pricing

Midcentury is sold on custom quotes, so its prices aren't published.

Prices can change and may vary by country or exclude tax. Check the official pricing page.

Midcentury alternatives

The main Midcentury alternatives listed on Debuly are Galileo, Cortex AI, Rescale, runAtlas, Alibaba Cloud Model Studio and Encord. The table compares pricing and platforms; order is A–Z.

1. 

   Alibaba Cloud Model Studio

   Alibaba Cloud platform giving developers API access to Qwen and partner models for text, image, video and speech, billed per token or by subscription.

   Developer platformsAudio and voiceVideo
   PaidVisit Alibaba Cloud Model Studio (official site)
2. 

   Cortex AI

   Real-world egocentric and robot training data for embodied AI, plus a beta platform for steering robots with AI models.

   Developer platformsAutomation and agents
   Visit Cortex AI (official site)
3. 

   Encord

   Multimodal data platform for AI teams to curate, annotate and evaluate image, video, DICOM, audio, document and text data.

   Developer platformsAutomation and agentsResearch and search
   Contact salesVisit Encord (official site)
4. 

   Galileo

   AI observability and eval platform that tunes evals to your data and turns them into guardrails for LLM apps and agents.

   Developer platformsAutomation and agents
   FreemiumVisit Galileo (official site)
5. 

   Rescale

   Rescale is a cloud platform for high-performance computing, simulation, AI physics and agentic engineering, sold as quoted enterprise editions.

   Developer platformsAutomation and agents
   Contact salesVisit Rescale (official site)
6. 

   runAtlas

   Enterprise AI stack with an OpenAI-compatible inference API, voice models, agent runtime and a security gateway.

   Developer platformsAutomation and agentsAudio and voice
   FreeVisit runAtlas (official site)

Midcentury FAQ

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

Image credits

• Logo: from midcentury.xyz · fetched 24 Oct 2026 · Vendor trademark, used only to identify the product; removed on request
• Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.