- Published on
Can a 2.6B parameter model outperform models ten to thirty times its size?
Benchmarks
All Posts
- update (14)
- openai (12)
- anthropic (8)
benchmarks (8)
- rag (4)
- google (4)
- how-to (3)
- visualization (3)
- agents (3)
- webgpu (3)
- wasm (3)
- transformersjs (3)
- local-ai (3)
- ml (2)
- short_read (2)
- llm (2)
- onnx (2)
- python (2)
- retrieval (2)
- open-source (2)
- maths (1)
- news (1)
- sovereign-ai (1)
- data-center (1)
- compute (1)
- india (1)
- gpu (1)
- infrastructure (1)
- ai-policy (1)
- browser-ai (1)
- pglite (1)
- webllm (1)
- pgvector (1)
- mlx (1)
- sft (1)
- reasoning (1)
- qwen (1)
- deepseek (1)
- voice (1)
- markdown (1)
- pdf (1)
- typesetting (1)
- serverless (1)
- aws (1)
- azure (1)
- gcp (1)
- sandboxing (1)
- industry (1)
- Published on
Elon Musk earlier wrote that source code is on the verge of becoming like assembly, and that the next step is generating efficient binaries directly with AI ...- Published on
OpenAI released GPT-6 Astra this week, and it is a generational leap. Most attention is focused on its reasoning variants, which set new records across the b...- Published on
Meta released Muse Spark 1.3 earlier this week. On the Artificial Analysis Intelligence Index, it sits among the leading models, level with Claude Fable 5 an...- Published on
GPT-6 Astra scores 61 on the Artificial Analysis Intelligence Index. That is the same score as GPT-5.6 Sol. Many people are looking at the composite benchmar...- Published on
GPT-6 Astra scores 62.7% on ARC-AGI-3 Semi-Private under the Standard harness. It also surpasses the human baseline in action efficiency on the vast majority...- Published on
Claude Fable 5.1 becomes the first model to beat human baseline score on SimpleBench with Gemini 3.8 Flash very close behind.- Published on
This is very interesting. New release from Google DeepMind, Gemini 3.8 Flash now leads the DeepSWE benchmark (widely considered the most reliable benchmark f...