---
title: "AlphaSignal Benchmarks News"
url: "https://alphasignal.ai/topic/benchmarks"
description: "Latest AI and machine learning news about Benchmarks on AlphaSignal."
---

# AlphaSignal Benchmarks News

Latest AI and machine learning news about Benchmarks on AlphaSignal.

## Articles

- [China Telecom's Xing4.0 Runs a 29B Coding Agent on 19GB Locally](https://alphasignal.ai/news/china-telecom-s-xing4-0-runs-a-29b-coding-agent-on-19gb-locally.md) (2026-10-04)
- [livenerf Catches Claude Opus 5.5 Drifting With a Daily 78-Question Trap](https://alphasignal.ai/news/livenerf-catches-claude-opus-5-5-drifting-with-a-daily-78-question-trap.md) (2026-10-03)
- [Ideogram 4.5 Fights Edit Drift With 27x Pricing Spread and Multi-Turn Precision](https://alphasignal.ai/news/ideogram-4-5-fights-edit-drift-with-27x-pricing-spread-and-multi-turn-precision.md) (2026-10-03)
- [Epoch Tracked 8.3 Million ChatGPT Messages to Reveal How Users Deepened Their Habits](https://alphasignal.ai/news/epoch-tracked-8-3-million-chatgpt-messages-to-reveal-how-users-deepened-their.md) (2026-10-01)
- [Stanford's ScholarCatalyst Reveals AI Misses 52% of Key Research Papers](https://alphasignal.ai/news/stanford-s-scholarcatalyst-reveals-ai-misses-52-of-key-research-papers.md) (2026-10-01)
- [When to use DeepSeek-V4.1-Flash](https://alphasignal.ai/news/when-to-use-deepseek-v4-1-flash.md) (2026-10-01)
- [Microsoft's MAI-Transcribe-2-Streaming Tops 38 Models With 2.5% Error Rate](https://alphasignal.ai/news/microsoft-s-mai-transcribe-2-streaming-tops-38-models-with-2-5-error-rate.md) (2026-10-01)
- [Video-Index Exposes 35 Video Benchmarks AI Can Ace Without Watching](https://alphasignal.ai/news/video-index-exposes-35-video-benchmarks-ai-can-ace-without-watching.md) (2026-10-01)
- [Upstage's Solar Mini 4 Triples Its Benchmark Score With 3B Active Parameters](https://alphasignal.ai/news/upstage-s-solar-mini-4-triples-its-benchmark-score-with-3b-active-parameters.md) (2026-10-01)
- [Google's Gemini 4 Argon Tops Vals Index at 68.9% Using Fewer Tokens](https://alphasignal.ai/news/google-s-gemini-4-argon-tops-vals-index-at-68-9-using-fewer-tokens.md) (2026-09-30)
- [Perplexity's pplx-embed-v2 Beats Voyage at 8x Smaller Vector Size](https://alphasignal.ai/news/perplexity-s-pplx-embed-v2-beats-voyage-at-8x-smaller-vector-size.md) (2026-09-30)
- [EurekaBench Reveals AI Agents Still Fall 27 Points Behind Humans on Real Science](https://alphasignal.ai/news/eurekabench-reveals-ai-agents-still-fall-27-points-behind-humans-on-real-science.md) (2026-09-30)
- [Utopai Studios' Utopai X Beats the MiniMax H3 Model It Was Built On](https://alphasignal.ai/news/utopai-studios-utopai-x-beats-the-minimax-h3-model-it-was-built-on.md) (2026-09-30)
- [Artificial Analysis Rebuilds AA-Video to Rank Models by Real-World Jobs](https://alphasignal.ai/news/artificial-analysis-rebuilds-aa-video-to-rank-models-by-real-world-jobs.md) (2026-09-30)
- [Tencent's ExplorationBench Forces AI to Discover Hidden Rules Through Active Experimentation](https://alphasignal.ai/news/tencent-s-explorationbench-forces-ai-to-discover-hidden-rules-through-active.md) (2026-09-30)
- [Cohere's Embed 5 Lets Developers Index Smart but Query 2.4x Faster](https://alphasignal.ai/news/cohere-s-embed-5-lets-developers-index-smart-but-query-2-4x-faster.md) (2026-09-30)
- [Hugging Face's ELECTRA Reranker Adds ONNX and OpenVINO for Faster CPU Search](https://alphasignal.ai/news/hugging-face-s-electra-reranker-adds-onnx-and-openvino-for-faster-cpu-search.md) (2026-09-30)
- [Vals AI Rebuilds Vals Index 2.1 Adding Tax and Overhauled Coding Scores](https://alphasignal.ai/news/vals-ai-rebuilds-vals-index-2-1-adding-tax-and-overhauled-coding-scores.md) (2026-09-29)
- [Cerebras Beats Human Speed, Booking Restaurants 19x Faster Than Rivals](https://alphasignal.ai/news/cerebras-beats-human-speed-booking-restaurants-19x-faster-than-rivals.md) (2026-09-29)
- [DeepSeek V4 Flash Slashes AI Reasoning Costs by 99% in 19 Months](https://alphasignal.ai/news/deepseek-v4-flash-slashes-ai-reasoning-costs-by-99-in-19-months.md) (2026-09-29)
- [TabFM Beats AutoML on 51 Datasets Without a Single Training Update](https://alphasignal.ai/news/tabfm-beats-automl-on-51-datasets-without-a-single-training-update.md) (2026-09-29)
- [LibraryDesignBench Catches AI Coding Agents Rebuilding Code That Already Exists](https://alphasignal.ai/news/librarydesignbench-catches-ai-coding-agents-rebuilding-code-that-already-exists.md) (2026-09-29)
- [Vals AI's Ten Claude Agents Formally Prove a Century-Old Math Problem](https://alphasignal.ai/news/vals-ai-s-ten-claude-agents-formally-prove-a-century-old-math-problem.md) (2026-09-29)
- [Vals AI's MysteryMechanism Benchmark Shows GPT-6 Astra Tops Science Test at 53%](https://alphasignal.ai/news/vals-ai-s-mysterymechanism-benchmark-shows-gpt-6-astra-tops-science-test-at-53.md) (2026-09-28)
- [Anthropic's Claude Sonnet 5.5 Jumps 60 Points in Agentic Coding](https://alphasignal.ai/news/anthropic-s-claude-sonnet-5-5-jumps-60-points-in-agentic-coding.md) (2026-09-28)
- [Fish Audio's transcribe-1-pro Adds Speaker ID and Emotions for $0.36 an Hour](https://alphasignal.ai/news/fish-audio-s-transcribe-1-pro-adds-speaker-id-and-emotions-for-0-36-an-hour.md) (2026-09-28)
- [ElevenLabs' Eleven v4 Tops English Voice AI With 91.7% Pronunciation Score](https://alphasignal.ai/news/elevenlabs-eleven-v4-tops-english-voice-ai-with-91-7-pronunciation-score.md) (2026-09-28)
- [Artificial Analysis' Cyber Index Exposes a 65x Cost Gap Among Top AI Models](https://alphasignal.ai/news/artificial-analysis-cyber-index-exposes-a-65x-cost-gap-among-top-ai-models.md) (2026-09-28)
- [GPT, Gemini, and Claude All Pick the Same Right Answers](https://alphasignal.ai/news/gpt-gemini-and-claude-all-pick-the-same-right-answers.md) (2026-09-28)
- [Jina AI's jina-reranker-v3 Ranks 64 Documents Together in One Pass](https://alphasignal.ai/news/jina-ai-s-jina-reranker-v3-ranks-64-documents-together-in-one-pass.md) (2026-09-27)
- [Zed's Delta Beats Codex and Claude Code at Half the Requests](https://alphasignal.ai/news/zed-s-delta-beats-codex-and-claude-code-at-half-the-requests.md) (2026-09-25)
- [Sarvam Releases Saaras V4 to Transcribe All 22 Indian Languages at Once](https://alphasignal.ai/news/sarvam-releases-saaras-v4-to-transcribe-all-22-indian-languages-at-once.md) (2026-09-25)
- [Ornith 1.5 Beats 35B Models on Reasoning With Just 9B Parameters](https://alphasignal.ai/news/ornith-1-5-beats-35b-models-on-reasoning-with-just-9b-parameters.md) (2026-09-25)
- [Stanford's Terminal-Bench-Science Exposes GPT-6 Astra Leading at 63% on Real Science Tasks](https://alphasignal.ai/news/stanford-s-terminal-bench-science-exposes-gpt-6-astra-leading-at-63-on-real.md) (2026-09-24)
- [LM Arena Merges 12 AI Leaderboards Into One Free Dashboard](https://alphasignal.ai/news/lm-arena-merges-12-ai-leaderboards-into-one-free-dashboard.md) (2026-09-24)
- [CMU Builds a Synthetic Hospital Where AI Fails 45% of Medical Summaries](https://alphasignal.ai/news/cmu-builds-a-synthetic-hospital-where-ai-fails-45-of-medical-summaries.md) (2026-09-24)
- [Sarvam AI's Vision 2.1 Beats Gemini at Reading India's 22 Languages](https://alphasignal.ai/news/sarvam-ai-s-vision-2-1-beats-gemini-at-reading-india-s-22-languages.md) (2026-09-24)
- [Penn Researchers Show Jev Cuts AI Grading Costs by 325x](https://alphasignal.ai/news/penn-researchers-show-jev-cuts-ai-grading-costs-by-325x.md) (2026-09-24)
- [UT Austin Reveals Why Efficient AI Models Fail at 10M-Token Reasoning](https://alphasignal.ai/news/ut-austin-reveals-why-efficient-ai-models-fail-at-10m-token-reasoning.md) (2026-09-24)
- [Anthropic's Claude Opus 5.5 Tops Coding Benchmarks but Costs $13 per Task](https://alphasignal.ai/news/anthropic-s-claude-opus-5-5-tops-coding-benchmarks-but-costs-13-per-task.md) (2026-09-24)
- [OpenAI's MentalHealthBench Tests AI on Everyday Stress Beyond Crisis Responses](https://alphasignal.ai/news/openai-s-mentalhealthbench-tests-ai-on-everyday-stress-beyond-crisis-responses.md) (2026-09-23)
- [StudentBench Shows AI Tutors Match Human GRE Gains at 918x Lower Cost](https://alphasignal.ai/news/studentbench-shows-ai-tutors-match-human-gre-gains-at-918x-lower-cost.md) (2026-09-23)
- [Epoch's FAB Benchmark Watched GPT-6 Astra Jump from 28% to 80%](https://alphasignal.ai/news/epoch-s-fab-benchmark-watched-gpt-6-astra-jump-from-28-to-80.md) (2026-09-23)
- [Microsoft's MAI-Image-2.6 Climbs to No. 2 Faster Than Any Rival](https://alphasignal.ai/news/microsoft-s-mai-image-2-6-climbs-to-no-2-faster-than-any-rival.md) (2026-09-23)
- [Alibaba's Qwen Intelligence Splits Phone Automation Across Three Specialized Agents](https://alphasignal.ai/news/alibaba-s-qwen-intelligence-splits-phone-automation-across-three-specialized.md) (2026-09-23)
- [Cartesia's Sonic 3.6 Dominates Eight of Nine Global Voice AI Leaderboards](https://alphasignal.ai/news/cartesia-s-sonic-3-6-dominates-eight-of-nine-global-voice-ai-leaderboards.md) (2026-09-22)
- [StepFun's StepAudio 3 ASR Tops Speech Benchmark With 1.7% Error Rate](https://alphasignal.ai/news/stepfun-s-stepaudio-3-asr-tops-speech-benchmark-with-1-7-error-rate.md) (2026-09-22)
- [New Paper Cracks the Tightest Possible AI Preference Learning Bound](https://alphasignal.ai/news/new-paper-cracks-the-tightest-possible-ai-preference-learning-bound.md) (2026-09-22)
- [DIFTA-3D Uses Meta's Frozen DINOv3 to Boost Indoor 3D Detection](https://alphasignal.ai/news/difta-3d-uses-meta-s-frozen-dinov3-to-boost-indoor-3d-detection.md) (2026-09-22)
- [CMU Finds JEV Matches GPT-6 Accuracy at 277 Times Lower Cost](https://alphasignal.ai/news/cmu-finds-jev-matches-gpt-6-accuracy-at-277-times-lower-cost.md) (2026-09-22)