#1 API for training & evaluating multimodal models. Get 5K+ human preferences/min to enable online RL, refine model behaviour, and evaluate outputs.
Latent Space
- Tweets 71
- Following 0
- Followers 359
- Likes 63
Pinned Tweet
today, we’re releasing the largest open-source human image preferences dataset, along with a $1 million data grant
- 2M+ annotations by real people
- 30 SOTA image models ranked
- 10 categories (marketing, product design, anime etc)
dataset + benchmark + grant details below:
29
57
445
30,035
Our latest model continues to top benchmarks:
Sonic-3.6 is #1 on @datapointai's TTS bench across all eight customer support categories including empathy, spell-outs, repairs and disfluency.
Ranked by 42K blind votes: trydatapoint.com/benchmark/l…
today, we're releasing the largest open-source human audio preferences dataset, focused on the customer support use-case
- 300K+ annotations by real people
- 15 SOTA TTS models ranked (Sonic 3.6, Grok TTS, Simba 3.2, Eleven Labs v3)
- 8 categories (IVR menus, empathy, escalations, refunds etc)
dataset + benchmark + frontier plot below:
5
7
62
5,269
Datapoint AI retweeted
Simba 3.2 by Speechify AI ranks as by far the highest value TTS model right now. At an ELO of 1080, Simba by Speechify is 40 points above Grok and Gemini, and 100 points above OpenAI. It’s also 65 points above ElevenLabs while being more than 10x more affordable.
Simba 3.2 by @SpeechifyAI is the best value TTS model right now.
It ranks #2 overall in our human preference benchmark while costing just $10/1M characters, putting it in a league of its own.
5
10
2,206
today, we're releasing the largest open-source human audio preferences dataset, focused on the customer support use-case
- 300K+ annotations by real people
- 15 SOTA TTS models ranked (Sonic 3.6, Grok TTS, Simba 3.2, Eleven Labs v3)
- 8 categories (IVR menus, empathy, escalations, refunds etc)
dataset + benchmark + frontier plot below:
14
40
418
31,329
the full dataset is free and open: huggingface.co/datasets/data…
overall and category-wise rankings of 15 TTS models are available on our TTS bench: trydatapoint.com/benchmark/l…
1
2
18
2,587
Simba 3.2 by @SpeechifyAI is the best value TTS model right now.
It ranks #2 overall in our human preference benchmark while costing just $10/1M characters, putting it in a league of its own.
2
15
37
4,482
P-Image-Ideogram, built with @ideogram_ai, now has a Very High mode.
Very High uses test-time scaling to push final image quality further. Built for hero assets, difficult prompts, and quality-critical generations when people and products have to pop.
• ~5.55s at 1K · $0.033/image (1K) · $0.066/image (2K)
• Five modes from Very Low ($0.003) through Very High
Again, benchmarks agree: This new very high mode is on efficiency-quality based on human and automated evals from @datapointai, @DesignArena, and P-Judger. Full benchmarks results follow in a later post on P-Bench.
Available on @cloudflare @ComfyUI @GammaApp @inference_sh @kittldesign @LeonardoAi @lovart_ai @magnific @picsart @replicate @runware @scenario_gg @togethercompute @wavespeed_ai @wiroai
👉 Free Playground: playground.pruna.ai/p-image-…
🧩 API: docs.pruna.ai/
📚 Model page: pruna.ai/p-image-ideogram
8
10
66
10,343
today, we’re releasing the largest open-source human image preferences dataset, along with a $1 million data grant
- 2M+ annotations by real people
- 30 SOTA image models ranked
- 10 categories (marketing, product design, anime etc)
dataset + benchmark + grant details below:
29
57
445
30,035
the full dataset is free and open: huggingface.co/datasets/data…
overall and category-wise rankings of 30 image models are available on our image bench: trydatapoint.com/benchmark/l…
4
2
31
3,358
academics and non-profits can apply for $1M in Datapoint credits to collect human preference data for image, video, audio and other multimodal research:
trydatapoint.com/grants/
1
11
1,175
Introducing HumanEvals: open source library for adding real human judgment to your multimodal model evals.
Send image, video, or audio outputs.
Get pairwise preferences, ratings, or rankings from real people in seconds.
github.com/impel-intelligenc…
4
5
27
4,852
Heavily inspired from AutoEvals by @braintrust
Scores come back AutoEvals compatible, so it drops into your existing eval pipeline!
3
232
Excited to have helped @PrunaAI collect 1M+ votes for image preference data in a very short time :~)
P-Image-Ideogram dominate the speed-quality and price-quality Pareto frontiers for image generation. It is the result of a unique collaboration with @ideogram_ai.
- Four modes (Very low, low, medium, high) for 1K-2K image generation.
- Optimal quality-efficiency with 0.4s-7.5s latency, and $0.003-$0.03 price.
- Structured JSON control & exact color control.
Available via our inference partners @Replicate @inference_sh @scenario_gg @wavespeed_ai @wiroai @magnific @prodialabs @togethercompute @runware @lovart_ai @ComfyUI @LeonardoAi @kittldesign @GammaApp @Picsart @TellersAI @Cloudflare Validated by our benchmark partners @DesignArena, @datapointai, @RapidataAI
👉 Try it on the playground for free: buff.ly/QmnA5P5
🧩 Sign in on the API: buff.ly/0iaZy8g
📚 Model page: buff.ly/XRaKSb0
3
6
26
7,390
We are now opening up API access for 5K+ high-quality annotations/minute to enable RLHF, refine model behavior, and evaluate outputs so you can produce the best multimodal models.
More here: trydatapoint.com
3
215
Human votes with @datapointai places P-Image-Ideogram as optimal on quality-speed and quality-price.
1
5
17
307
P-Image-Ideogram dominate the speed-quality and price-quality Pareto frontiers for image generation. It is the result of a unique collaboration with @ideogram_ai.
- Four modes (Very low, low, medium, high) for 1K-2K image generation.
- Optimal quality-efficiency with 0.4s-7.5s latency, and $0.003-$0.03 price.
- Structured JSON control & exact color control.
Available via our inference partners @Replicate @inference_sh @scenario_gg @wavespeed_ai @wiroai @magnific @prodialabs @togethercompute @runware @lovart_ai @ComfyUI @LeonardoAi @kittldesign @GammaApp @Picsart @TellersAI @Cloudflare Validated by our benchmark partners @DesignArena, @datapointai, @RapidataAI
👉 Try it on the playground for free: buff.ly/QmnA5P5
🧩 Sign in on the API: buff.ly/0iaZy8g
📚 Model page: buff.ly/XRaKSb0
19
33
130
21,564
Claude Code for talking to your customers
Today, we’re opening up Datapoint AI for anyone to use. It is by far the fastest way to understand what your customers want.
Type a question. Real people answer. You get a report back in ~10 minutes, not three weeks, and at a fraction of the cost.
1
1
10
447