World Wire

Anthropic’s Frontier Red Team has released FlashProve , an open-source research prototype for provable inference . FlashProve is a zkSNARK system from Anthropic that proves a reply came from a committed model. The proof reveals nothing about the model weights. On Llama 4 Maverick , a 400B mixture-of-experts model, it proves about 110 tokens/second on 1 H100. For developers, this targets a real gap: today you cannot check which model actually answered your API call. TL;DR Size: Proves a 400B MoE model (~17B active per token); context capped at 1,024 tokens per request. Runs on: Prover needs 1 H100 80 GB, 24 CPU cores and ~290 GB host RAM. Verifier runs on a CPU, no GPU. Performance: Orders of magnitude faster than prior open-source zkML systems on the same 400B model. Best: 110 tokens/second proving on 1 H100, with 18.78 MB proofs checked in 17 seconds. Worst: Proving costs ~70 to 95x the GPU time of plain vLLM inference. What problem does provab...

Soniox has released Soniox STT v6 , the latest generation of its speech-to-text API. Soniox STT v6 is a speech-to-text model that reads what a voice agent says while hearing the caller. It uses that text to recognize short replies and to drop the agent’s own voice. The release ships as stt-rt-v6 for streaming and stt-async-v6 for files , across 60+ languages. TL;DR Size: Soniox STT v6 is a proprietary hosted model, no parameter count published. Runs on: Soniox cloud API only (WebSocket streaming and async REST). No self-hosted or on-device option disclosed. Performance: Soniox calls it “significantly” more accurate than v5, but publishes no v6 numbers yet. Best: Price. $0.12/hour streaming and $0.10/hour async, with diarization and language ID bundled. Worst: Zero independent v6 results. Its v5 predecessor scored 1.34% semantic WER on Pipecat, behind v4’s 1.25%. What problem does Soniox STT v6 solve? In a cascaded voice a...

OpenAI has introduced Pando , a new Git backend that now serves all development in its monorepo. The OpenAI Pando Git backend keeps the standard Git client and protocol. Engineers and Codex agents still run git push and git fetch as before. Only the server changed. OpenAI team reports median push latency fell nearly 80%. Pando now delivers nearly a petabyte of Git data per day. TL;DR Scale: Nearly 1 PB of Git data per day; peak above 57 TB in 1 hour. Runs on: Commodity hardware; standard cloud VMs saturate 10 Gbps links on clone and fetch. Performance: Thousands of concurrent pushers without degradation, per OpenAI’s own testing. Best: Median push latency down nearly 80% (about 5x). Worst: Not open source; repeat branch pushes resend non-trunk history (p99 pack under 3 MiB). What is OpenAI Pando? Pando is a Git-compatible server backend from OpenAI that scales concurrent pushes and fetches for a monorepo. It was written in Rust and reache...

Zyphra has released Zyphra Hierarchical Attention (ZHA) , a sparse attention method built around a tree-structured indexer. ZHA targets a known gap in sparse attention: the indexer that picks tokens still scans the whole context. Zyphra’s technical paper shows ZHA can be distilled into frozen pretrained models with about 4B tokens. On Qwen3.6-35B-A3B it stays close to dense attention on reasoning, agentic coding and 256K retrieval benchmarks. TL;DR Size: It is a method that adds 81,920 trainable parameters to Qwen3.6-35B-A3B (about 0.00024%). Tested up to 256K context. Runs on: Retrofits frozen pretrained models after a 4B-token KL warmup at 32K context. Speedups come from a modeled H100 BF16 roofline. No optimized kernels released. An FP4 router variant was also tested. Performance: Within a few points of dense attention on 7 reasoning, coding and retrieval benchmarks. Best: LongBench v2 hits 63.4% (deep tree) vs 58.0% dense, using about 1.1% of keys. ...