repoGitHubTrust 82 · PrimaryPublished 1mo agoLive · yesterday
SemiAnalysisAI/InferenceX
Open Source Continuous Inference Benchmark Research Platform — Kimi K3 2.8T, MiniMax M3, DeepSeekv4, GLM5 - GB200 NVL72 vs MI355X vs B200 vs GB300 NVL72 & soon™ TPUv6e/v7/Trainium2/3 | 开源持续推理基准研究平台 — Kimi K2.7-Code、MiniMax M3、DeepSeekv4、GLM5 - GB200 NVL72 vs MI355X vs B200 vs GB300 NVL72，即将推出™ TPUv6e/v7/Trainium2/3
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- PossiblePossibly related (embedding) · 53%OpenAI and Broadcom unveil LLM-optimized inference chip →
- PossiblePossibly related (embedding) · 49%Is it agentic enough? Benchmarking open models on your own tooling →
- PossiblePossibly related (embedding) · 48%Hardware startup unveils inference accelerator →
- PossiblePossibly related (embedding) · 48%GLM-5.2: Built for Long-Horizon Tasks →
- PossiblePossibly related (embedding) · 48%[Benchmark] Kimi K2.7 Code Q3 on Mac Studio M3 Ultra + RTX PRO 6000 over llama.cpp RPC: prefill improves, no changes in token generation/decode →
- PossiblePossibly related (embedding) · 47%Local benchmarks with a RTX 3090 - Qwen3.6 27b vs Ornith →
- PossiblePossibly related (embedding) · 47%GLM5.2 on AMD MI355X at 2626 tok/s/node at over 2x lower cost than Blackwell →
- PossiblePossibly related (embedding) · 49%Ran a classic(medival europe) fantasy RP/agentic benchmark across 8 local models Qwen3.6-27B held up better than its size suggests →
Covers
newsOpenAI and Broadcom unveil LLM-optimized inference chipnewsIs it agentic enough? Benchmarking open models on your own toolingnewsHardware startup unveils inference acceleratornewsGLM-5.2: Built for Long-Horizon Tasksnews[Benchmark] Kimi K2.7 Code Q3 on Mac Studio M3 Ultra + RTX PRO 6000 over llama.cpp RPC: prefill improves, no changes in token generation/decode
Covers (incoming)
newsLocal benchmarks with a RTX 3090 - Qwen3.6 27b vs OrnithnewsGLM5.2 on AMD MI355X at 2626 tok/s/node at over 2x lower cost than BlackwellnewsRan a classic(medival europe) fantasy RP/agentic benchmark across 8 local models Qwen3.6-27B held up better than its size suggestsnewsGLM-5.2 vs DeepSeek V4 vs Kimi K2.6: 62% SWE Pro [2026] - tech-insider.orgnewsQwen 3.6 27B - VLLM Performance Benchmark Results (BF16, FP8, NVFP4)newsBest Local VLMs - July 2026newsCPU TTS benchmark with UTMOS MOS scoring: Kokoro, Supertonic, Inflect-Nano, and Kyutai's new Pocket TTS [P]newsTencent has released its AI model 'Hy3' as an open model, claiming it is comparable to GLM-5.2 and DeepSeek-V4 at 295B and surpasses GPT-5.5 in scientific tasks. - GIGAZINEnewsBenchmarks compare open models against closed products, not closed models. We might be missing what were actually paying fornewsGLM-5.2 on 8xB200: the deployment math nobody spells out - NVFP4 + 2x TP=4 replicas should beat TP=8 by ~2x. Full config guidance inside.newsAccording to DataBricks, pi-coding-agent is ~2x cheaper than CC/Codex, GLM 5.2 on par with Opus 4.8 highnewsHy3 (295B MoE) and NVIDIA Nemotron-Labs-Audex-30B-A3B (audio-capable 30B MoE) GGUF quantsnewsGLM 5.2 running on MacBook Pro M5 48 GB Ram at between 2 - 2.8t/snewsCloud-vLLM Benchmark Differences [R]newsKimi K3 in the next few hours. Deepseek V4 GA later in the week. New Liquid models. New Mistral models sometime this month. And some rumours suggest GLM 5.5 is coming in August. Openweight AI is eating good.newsPrism-ML's Bonsai-27B BenchmarksnewsKimi K3 Intelligence, Performance and Price AnalysisnewsKimi K3 vs DeepSeek V4 Pro vs GLM-5.2: Open Trillion-Scale MoE Models Compared on Benchmarks, License, and Serving Cost - MarkTechPostnewsShow HN: Echo – Fable-level results at 1/3 the cost using open-weight modelsnewsCPU-only inference on a Celeron N5095 SBC: 6 models from 0.6B to 8B, benchmarked
Related across the graph
newsRan a classic(medival europe) fantasy RP/agentic benchmark across 8 local models Qwen3.6-27B held up better than its size suggestsnewsGLM-5.2 vs DeepSeek V4 vs Kimi K2.6: 62% SWE Pro [2026] - tech-insider.orgnewsCloud-vLLM Benchmark Differences [R]newsKimi K3 Intelligence, Performance and Price AnalysisnewsGLM-5.2: Built for Long-Horizon TasksnewsQwen 3.6 27B - VLLM Performance Benchmark Results (BF16, FP8, NVFP4)newsGLM 5.2 running on MacBook Pro M5 48 GB Ram at between 2 - 2.8t/snewsOpenAI and Broadcom unveil LLM-optimized inference chipnewsKimi K3 vs DeepSeek V4 Pro vs GLM-5.2: Open Trillion-Scale MoE Models Compared on Benchmarks, License, and Serving Cost - MarkTechPostnewsBest Local VLMs - July 2026newsHy3 (295B MoE) and NVIDIA Nemotron-Labs-Audex-30B-A3B (audio-capable 30B MoE) GGUF quantsnewsGLM-5.2 on 8xB200: the deployment math nobody spells out - NVFP4 + 2x TP=4 replicas should beat TP=8 by ~2x. Full config guidance inside.newsPrism-ML's Bonsai-27B BenchmarksnewsCPU TTS benchmark with UTMOS MOS scoring: Kokoro, Supertonic, Inflect-Nano, and Kyutai's new Pocket TTS [P]newsKimi K3 in the next few hours. Deepseek V4 GA later in the week. New Liquid models. New Mistral models sometime this month. And some rumours suggest GLM 5.5 is coming in August. Openweight AI is eating good.newsTencent has released its AI model 'Hy3' as an open model, claiming it is comparable to GLM-5.2 and DeepSeek-V4 at 295B and surpasses GPT-5.5 in scientific tasks. - GIGAZINEnews[Benchmark] Kimi K2.7 Code Q3 on Mac Studio M3 Ultra + RTX PRO 6000 over llama.cpp RPC: prefill improves, no changes in token generation/decodenewsGLM5.2 on AMD MI355X at 2626 tok/s/node at over 2x lower cost than BlackwellnewsShow HN: Echo – Fable-level results at 1/3 the cost using open-weight modelsnewsCPU-only inference on a Celeron N5095 SBC: 6 models from 0.6B to 8B, benchmarkednewsBenchmarks compare open models against closed products, not closed models. We might be missing what were actually paying fornewsAccording to DataBricks, pi-coding-agent is ~2x cheaper than CC/Codex, GLM 5.2 on par with Opus 4.8 highnewsHardware startup unveils inference acceleratornewsIs it agentic enough? Benchmarking open models on your own toolingnewsLocal benchmarks with a RTX 3090 - Qwen3.6 27b vs Ornith