Reka
@RekaAILabs
Followers
16K
Following
523
Media
27
Statuses
153
An AI research and product company 🫠. We are a team of scientists and engineers building state-of-the-art multimodal reasoning models 😻
Joined June 2022
⚡We are open sourcing Reka Flash 3, our new reasoning model that was trained from scratch. It achieves competitive performance with only 21B parameters. ⚡. Reka Flash 3 powers Nexus, our new enterprise intelligence platform. 🤖
54
125
666
Vision capabilities are core to our mission to develop multimodal AI that understands both the digital and physical worlds. In our latest blogpost, we discuss our approach to process video inputs and our industry-leading results from different domains.
6
18
65
RT @ArtificialAnlys: Reka AI has launched Reka Flash 3, a new open source 21B parameter reasoning model - the highest scoring model of its….
0
34
0
RT @TheXeophon: After multiple requests, here it finally is: @RekaAILabs Flash 3 is an amazing model on my vibe eval!. While its consistenc….
0
6
0
Reka Flash 3 is available via our API as reka-flash-3. See for details on how to use it.
1
3
30
RT @DaniYogatama: there shouldn't be any distinction between reasoning and non-reasoning models if the goal is evaluating intelligence. bot….
0
3
0
Reka Flash 3 is the best open-source model for agentic long-horizon tasks (on BALROG, third place overall, competitive with Sonnet-3.5 and 4o). We use a customized version to power Nexus, our enterprise agentic platform.
A new challenger has entered the ring 🥉. This week’s entry on takes third place, powered by a 21B reasoning model. @RekaAILabs Reka Flash 3 dominates similarly sized reasoning models like DeepSeek-R1-Distill-Qwen 32B on BALROG’s toughest agentic tasks!.🧵
4
13
80
Reka Flash 3 update: added another great model Gemma-3 27B (instruction tuned version) to our benchmark results since many people asked us how they compare on these datasets. Also Apache 2.0 vs Gemma license.
6
13
123
RT @RekaAILabs: ⚡We are open sourcing Reka Flash 3, our new reasoning model that was trained from scratch. It achieves competitive performa….
0
125
0
⚡We are open sourcing Reka Flash 3, our new reasoning model that was trained from scratch. It achieves competitive performance with only 21B parameters. ⚡. Reka Flash 3 powers Nexus, our new enterprise intelligence platform. 🤖
54
125
666
Reka Flash 3 was finetuned on synthetic and public datasets, followed by RLOO with model-based and rule-based rewards. 🏋️. We found that forcing the model to output </reasoning> is an effective way to control the tradeoff between quality and thinking time. 🤔. 🌐Blog:
1
10
76
We use Reka Flash 3 to power Nexus, our new platform for organizations to create and manage AI workers. 🤖🤝🧑⚕️🧑💼🕵️🧑🔬🧑🚀🟰🚀🚀🚀. Nexus workers have native deep research capabilities and can browse the web, execute code, and analyze internal files including documents, images,
2
16
72
RT @PalioAI: 𝐏𝐚𝐥𝐢𝐨 𝐱 𝐁𝐞𝐟𝐫𝐢𝐞𝐧𝐝. Having fun taming AI & earning $XTER in Befriend?. The LLM language model is supported by Xterio's AI partn….
0
128
0
🎉We are excited to introduce Space, a new AI-powered canvas that redefines how you interact with our multimodal models. Most of our interactions with AI today are carried out in a chat interface. Many of you have experienced our models via Reka Chat. As our models become more
23
35
132
🖼️Richer output: What used to be just a text snippet is now a longer document with interleaved text and images, a spreadsheet, or even a video file. Space allows you to view, and edit multimodal data more easily.
2
0
11
AI models should continuously learn from our interactions with them and become personalized. This includes learning what kind of tasks we use them for as well as how we interact with them. Having a consistent graphical user interface allows us to gather data more easily as well.
0
0
14
RT @XterioGames: Xterio AI Agent Program .with @RekaAILabs, as their first Gaming partner!. 🎖 Leading AI Innovation.✨ Unparalleled AI Exper….
0
23
0