Click any tag below to further narrow down your results
Links
The author builds an interactive atlas of Qwen3-1.7B's residual stream, tracking how about 1,000 question-and-answer examples across four languages and Python land in the model's internal vector space. The result is a view of language and topic structure as overlapping geometric sheets.
- The axes track language strongly, but topic structure persists after accounting for language, so the model encodes both what is discussed and how it is said.
- Code examples trace paths through partly shared regions with natural-language examples, suggesting a common geometric structure across very different text.
- The headline 17.6% figure is token-state variance under the Global PCA axes, not a measure of meaning or reasoning, and the author cautions that the sample is small, templated, and single-turn rather than real multi-turn conversation.
Microsoft released three new models for building voice agents: a streaming transcription model that produces text in 100ms, and two text-to-speech models optimized for speed and cost. Together they're designed to handle the full loop of hearing, understanding, deciding, and speaking in real-time conversations.
- MAI-Transcribe-2-Streaming ranks #1 for accuracy on Artificial Analysis and delivers partial transcripts in just over 100ms, letting voice agents start acting mid-sentence rather than waiting for speakers to finish.
- MAI-Voice-2.1-Flash generates 45 seconds of audio with 150ms end-to-end latency and costs 60% less than comparable models at $15 per 1M characters, making it practical for high-volume applications.
- Both voice models support 23 languages with a single voice maintaining native accents across all of them, plus voice cloning from a few seconds of audio with built-in consent guardrails against misuse.
Cohere released North Small Translate, a sparse mixture-of-experts model with 25 billion active parameters designed for machine translation across 50 languages. The model scores 83.60 on WMT26 benchmarks and is available on Hugging Face under a non-commercial license.
- The model uses a sparse MoE architecture with 218B total parameters but only activates 25B per token, reducing computational overhead compared to dense models of similar capability
- It supports 50 languages including major ones like English, Spanish, Mandarin, and Hindi, plus less common ones like Filipino and Maltese
- Multiple quantization options (BF16, FP8, NVFP4) let you run it on different hardware—from 8 H100s down to 2 H100s depending on precision, with production deployment using greedy decoding
Microsoft released MAI-Transcribe-2, a speech recognition model that claims to outperform competitors like OpenAI's Whisper and Google's Gemini on speed, accuracy, and cost. It handles 60 languages, includes speaker identification and word-level timestamps, and costs $0.10 per hour.
- 10x faster than GPT-Transcribe, 7x faster than Scribe v2, and 5x faster than Gemini 3.5 while maintaining higher accuracy across benchmarks
- Supports 60 languages with 5.2% average word-error rate on FLEURS benchmark, plus features like speaker diarization, keyword biasing, and code-switching for mixed-language conversations
- Priced at $0.10/hour as a limited-time launch offer, undercutting competitor pricing while maintaining competitive performance
Mistral OCR 4 extracts text from PDFs, DOCs and more while also returning bounding boxes, block types and per-word confidence. It supports 170 languages, runs in a single container for self-hosted deployments, and outperforms rivals on human and automated benchmarks.
- Mistral OCR 4 doesn't just extract text—it returns bounding boxes, block types (titles, tables, equations, signatures), and per-word confidence scores, and can be self-hosted in a single container for data-sovereignty needs
- Beat every tested competitor in human evaluations across 600+ documents/12+ languages with a 72% average win rate, and scored 85.20 on OlmOCRBench and 93.07 on OmniDocBench
- Roughly 4x faster than some enterprise OCR providers, and 8x cheaper with 17x lower latency than leading agentic document parsers on finance datasets with charts/figures
- Priced at $4 per 1,000 pages via API ($2 with batch discount) or $5 per 1,000 pages through the no-code Document AI interface
Datalab’s 4 billion-parameter Chandra OCR 2 outperforms GPT-4o and Gemini across independent and multilingual benchmarks, handling complex layouts, math notation, flowcharts and 90 languages with state-of-the-art accuracy. It’s available under Apache 2.0 code with a modified OpenRAIL-M license for weights, runs locally via HuggingFace or vLLM, and doubles throughput over its predecessor.
- Chandra OCR 2 (4B params, open-weight) scored 85.9% on olmOCR vs GPT-4o's 69.9%, and beat Gemini/GPT-5 Mini on multilingual benchmarks, with huge gains on South Asian scripts (Kannada +42.6, Malayalam +46.2, Telugu +39.1).
- It processes full pages in one pass rather than splitting into blocks, giving it an edge on tables, nested headers, checkboxes, handwritten math, and flowcharts exported as Mermaid diagrams.
- Despite shrinking from 9B to 4B parameters, throughput doubled to ~2 pages/sec on an H100 while accuracy improved.
- Code is Apache 2.0 and installable via pip/Docker, but weights use a modified OpenRAIL-M license requiring a paid commercial license for larger companies.
Google’s new Gemini 3.5 Live Translate model converts speech to speech in real time across more than 70 languages, preserving speakers’ intonation, pacing and pitch. It streams audio continuously with minimal delay and is available via the Gemini Live API, Google Meet preview, and Google Translate apps. TAGS: live-translation, speech-translation, multilingual, real-time-ai, gemini-models
- Gemini 3.5 Live Translate does real-time speech-to-speech translation across 70+ languages while preserving intonation, pacing and pitch, with only a few seconds of delay
- Google Meet's speech translation jumps from 5 languages to 70+, now covering over 2,000 language pairs instead of just English-based ones
- Grab is testing it across 10 million voice calls a month for drivers and riders
- All translated audio carries an invisible SynthID watermark to keep AI-generated speech traceable
Chandra OCR 2, a 4 billion-parameter model from Datalab, outperforms GPT-4o and Gemini on AllenAI’s olmOCR benchmark and a 90-language test while halving the model size. It preserves layout, reads complex tables and math notation, converts diagrams to Mermaid, and runs at two pages per second on an NVIDIA H100. The code is Apache 2.0 but the model weights use an OpenRAIL-M license with commercial restrictions.
- Chandra OCR 2 scores 85.9% on olmOCR vs GPT-4o's 69.9%, while cutting model size from 9B to 4B parameters and hitting ~2 pages/sec on an H100
- Multilingual performance beats Gemini 2.5 Flash and GPT-5 Mini, with 40-46 point gains on scripts like Kannada, Malayalam and Telugu over Chandra 1
- Weights carry an OpenRAIL-M license requiring a paid commercial license once a company exceeds $2M in funding or revenue, despite Apache 2.0 code
- Handwriting recognition remains weak, dropping to ~50.4% accuracy on complex forms despite strong printed-text and table/math handling
This newsletter announces an online Dynamic Healing group session scheduled for Wednesday, May 8. It provides subscription details and multi-language translation options.
- Page is a Mailchimp scaffolding hub with Subscribe, Past Issues, and RSS links, not actual meeting content
- Translate dropdown offers over sixty language options, from Arabic and Chinese to Swahili and Ukrainian
- No session details, agenda, names, or dates for the Dynamic Healing Group meeting appear on this page
This page lists subscription options and language translations for a newsletter. Users can choose from over 40 languages to read past issues via RSS or direct email.
- The actual article content (subscription/language options for a newsletter) is entirely mismatched with the detailed summary describing a Dynamic Healing Group session, suggesting a content extraction error
- The summary appears to be fabricated or pulled from a different source than the stated article