Opper retweeted
We're proud to see Aplomb 1 highlighted in @IBM Think as the most cost-efficient model @opperai runs on JevMan, their Pac-Man benchmark for decision models. Congratulations and big thanks to the Opper team!
Article: ibm.com/think/news/ai-decisi…
JevMan: opper.ai/jevman-benchmark/
Opper retweeted
The only benchmark you should definitely trust, and probably use, before moving your most critical production traffic to a decision model: JevMan
Because at the end of the day, what you really need to know is, can jev beat me at Pac-Man?
The answer is maybe: opper.ai/jevman-benchmark
Happy JevDay!
But before announcements we have to talk philosophy…
The needs of early adopters / tinkerers / dreamers are quite different from what is needed for extremely stable infrastructure.
For the Jev release, we erred on the latter, but this release is for the former - with a gate so people can opt-in to the danger zone!
Opper retweeted
Ha! This is great :D
ps: clef is powered by @sferenceai
opper.ai/jevman-benchmark/
Great to see @empiriolabsai new decision model Aplomb 1 on jevman, the Pac-Man benchmark for AI decision models. Check out their score, or try the model via API: opper.ai/empiriolabs/aplomb-…
Aplomb 1 is now on @opperai, the European AI gateway, along with our other natively hosted models.
On Opper's jevman Pac-Man benchmark, the 5B-parameter model made the top 10 and ranked #1 for cost among the models Opper ran itself.
opper.ai/jevman-benchmark/#l…
Opper retweeted
Aplomb 1 is now on @opperai, the European AI gateway, along with our other natively hosted models.
On Opper's jevman Pac-Man benchmark, the 5B-parameter model made the top 10 and ranked #1 for cost among the models Opper ran itself.
opper.ai/jevman-benchmark/#l…
Opper retweeted
Congrats @CompleteSkeptic and @typesafeai on the raise. We benchmarked jev on jevman, the Pac-Man benchmark for AI decision models, and it came out on top! opper.ai/jevman-benchmark/
launched a model
accidentally served trillions of tokens a day
29.4% of the fortune 500 showed up
raised a really big series A from @a16z
it's been 3 weeks
we would like to sleep now
Opper retweeted
Introducing jevman benchmark: AI decision models play Pac-Man. Luna ties for first, check the full results, or join the benchmark and play against Luna. Can you beat the high score?
opper.ai/jevman-benchmark/
New model Typesafe.ai/Jev is now live on Opper
Jev is a new kind of model from TypeSafe AI. It doesn't generate text. It generates decisions.
Give it state, ask it typed questions:
→ choice: pick one of N labels
→ score: rate against a rubric
→ noul: is this true?
Every answer comes back as a probability, not just a label. So your code can act when the model is sure, and escalate when it isn't. Built for fast calls inside software: triage, routing, moderation, agent guardrails etc. Responses come back in milliseconds, not seconds.
Pricing at $0.042 per million input tokens, with output free.
Available today through a new classifier endpoint on Opper, and it works with the TypeSafe SDK out of the box. Just point it at Opper, use your Opper key and make it conform to budget, auditing requirements and policies.
Congrats to the TypeSafe team on the launch 👏
🤖 Made with AI
Nextbit is live on Opper, the first Spanish route on our gateway. DeepSeek V4, Kimi K3, ALIA 40B and more, served from Spain. opper.ai/provider/nextbit
Describe a workflow in plain English and @realPUNKUAI builds and runs the agent. Now you can choose @opperai for the models behind it. Try it at: opper.ai/apps/punku
Mark Zuckerberg announced open weights for Meta's Muse models today. Muse Glimmer is out now and Muse Spark 1.2 will follow soon.
You don't have to wait for the weights. Muse Spark 1.2 is already live on Opper. It handles real software engineering work across large repos and powers Muse Code, Meta's new coding agent.
Try it today: opper.ai/meta/muse-spark-1-2
Qwen3.8 Max by @Alibaba_Qwen is live on Opper, Europe's AI gateway.
Alibaba's new flagship, built for long-horizon coding, research, and multimodal agent work.
Give it a try at: opper.ai/alibaba/qwen3-8-max
The cheapest frontier class model to run, DeepSeek V4 Flash (0731), is now live on Opper AI's EU endpoint 🇪🇺
opper.ai/deepseek/deepseek-v…
The strongest open-weight model, Kimi K3 by @Kimi_Moonshot, is now live on @opperai's EU endpoint. Take it for a walk without your data leaving the EU 🇪🇺
Launching @opperai on @ProductHunt today! We also launched Media Studio so you can create launch videos like this one.
🤖 Made with AI
