Perplexity launches Bumblebee: How its new read-only dev scanner differs from Chainguard
The AI company's Bumblebee tool tackles your most urgent question after any supply‑chain advisory: Do your programmers have this malware installed?
RADAR ARCHIVE
Week 22, 2026
10 tools added to the directory and 2 launches tracked, May 25 – 31, 2026.
Added to the directory (10)
Tested and written up.
Anthropic's powerful enterprise model from May 2026
Claude Opus 4
Why: Claude Opus 4.8 continues Anthropic's reputation for reliable, steerable models. It is a top choice for enterprises that need a capable assistant with strong safety characteristics and nuanced writing.
Enterprise
Best for Enterprise Reasoning
Visit
Mistral's unified work and coding agent
Mistral Vibe is Mistral's unified agent for work and coding, rebranded from Le Chat and launched on May 28, 2026
Why: Mistral Vibe brings Mistral's strong European model lineage into a competitive all-in-one agent. It is a good choice for users who want a privacy-conscious alternative to US-centric assistants with solid coding skills.
Freemium
Best for European AI Assistant
Visit
DeepSeek's open-weight model with permanent pricing
DeepSeek V4-Pro is a high-performance language model from DeepSeek
Why: DeepSeek V4-Pro stands out for combining frontier-level performance with transparent, permanent pricing and open weights. It is a practical choice for teams that want to self-host or avoid unpredictable API costs.
Freemium
Best for Predictable Pricing
Visit
MiniMax's 1M-context agentic frontier model
MiniMax M3 is a 1-million-token-context agentic frontier model released on May 31, 2026
Why: MiniMax M3's 1M context window makes it competitive for tasks that require digesting entire codebases, books, or video transcripts in a single pass. It is a strong option for long-context agentic applications.
Freemium
Best for 1M Context
Visit
StepFun's 198B MoE vision-language model
StepFun Step 3
Why: Step 3.7 Flash offers a competitive Chinese-frontier multimodal model with an MoE architecture that balances capability and inference cost. It is a useful option for vision-language applications and for teams exploring alternatives to US models.
Freemium
Best for Efficient VLM
Visit
AI music generation with professional controls
ElevenLabs Music v2, released on May 26, 2026, is the company's next-generation AI music generator
Why: Music v2 extends ElevenLabs' voice and audio strengths into complete song generation. For creators who already use ElevenLabs for voice, it offers a natural path to full music production.
Freemium
Best for AI Music Production
Visit
AI-powered video dubbing in multiple languages
ElevenLabs Dubbing v2, released on May 28, 2026, automatically translates and dubs video content into multiple languages while preserving the original speaker's voice characteristics and lip-sync timi...
Why: Dubbing v2 makes multilingual video production far more accessible. It is especially valuable for creators, educators, and businesses that want to localize content without hiring voice actors for every language.
Freemium
Best for AI Dubbing
Visit
Open general-purpose multimodal video generation model
MiniMax H3 is a next-generation open general-purpose multimodal video model that generates video from text prompts, images, first-last-frame references, and multimodal inputs
Why: MiniMax H3 is the provider's current flagship video model, replacing the Hailuo 2.x series with an open, higher-resolution multimodal generation pipeline.
Freemium
Best for Multimodal Video
Visit
Fast text-to-3D and image-to-3D generation
Tripo v3
Why: Tripo v3.1 refines one of the fastest production-ready 3D generators. It is a practical choice for game developers, product designers, and AR/VR creators who need usable assets quickly.
Freemium
Best for Fast 3D Assets
Visit
Hyper3D's image-to-3D generation model
Rodin Gen-2, also referred to as Rodin 2
Why: Rodin Gen-2 offers a focused image-to-3D pipeline that is easy to use for artists and developers. It is a solid option when you have a 2D concept and need a 3D starting point quickly.
Freemium
Best for Image-to-3D
Visit
On the radar (2)
Reported in the daily briefing this week. Not reviewed for the directory.
Taiwan chip stocks climb after Nvidia announces $150 billion spending plans
Mainland China-based chip giants such as Cambricon saw their shares tumble on Wednesday.