AI Briefing#small-language-models
The latest AI and developer news about #small-language-models, with the original source and a short summary.
The latest AI and developer news about #small-language-models, with the original source and a short summary.
The open-source model achieves SOTA tool-use performance for its size, scoring 45.0 on average compared to 38.8 for Qwen3.5-2B-Think.
The 2B parameter model achieves a 0.834 field-level F1 score, surpassing Claude Opus 5 by 1.8 points and GPT-5.6 Luna by 9.3 points.
The study found byte-level models required about 6x less text to reach parity with token-based models during initial training.
A frozen 16 MB vocabulary table matches or exceeds learned embeddings on key benchmarks while reducing model size to 46 MB and boosting CPU inference speed by five to six times.
The 3.66B model achieves a 55.4 logic composite score, effectively tying with Qwen3-8B and offering high-speed CPU inference.
The 3.66B parameter model scores 55.4 on a logic composite, matching Qwen3-8B at less than half the size.
Europe-based Desert Ant Labs has released 18 specialized small models, including audio and vision, that run on-device without cloud dependency.
The new small LLM Spark-X2.5, featuring a proprietary architecture, has been released, with the 4B model demonstrating performance comparable to Qwen 3.5 9B in benchmarks.
Artificial Analysis, in collaboration with Liquid AI, released a benchmark measuring the intelligence and inference speed of small LLMs under 8GB on the iPhone 17 Pro.
Luth-2 has been released as French-specialized 0.8B and 2B models.
Cactus has released Needle 2, a 14MB on-device agentic LLM.
A technique is proposed that separates logic and execution in small LLMs to produce deterministic results without token overhead.
mii-llm has released a small language model with about 400 million parameters supporting Portuguese and English.
Cactus added a 68k-parameter probe layer to Gemma 4 E2B and unveiled a hybrid inference system that uses the model's own error detection as a signal for cloud routing.
OpenBMB has released MiniCPM5-2B, a model that recorded top-tier performance among 4B-class models.
An ultra-lightweight 51M-parameter routing model that selects the appropriate model based on user requests has been released.
VibeThinker-3B, a small language model (SLM), has demonstrated frontier-level performance on math and coding benchmarks.
Mistral AI declared its transition into a full-stack AI company spanning compute and platform, beyond just model development.
A 3-billion-parameter specialized model outperformed all commercial frontier APIs, including Claude Opus and GPT-5.4, on an OCR benchmark.
SupraLabs has released Supra-50M, a small language model with 50M parameters trained on 20 billion tokens.
New research shows guardrail techniques can raise an 8B-scale model's agentic task success rate from 53% to 99%.
Sapient Intelligence has launched HRM-Text 1B, a 1B model that achieves high reasoning performance with little data and low cost.
mii-llm released the 0.4B multilingual models Zagreus/Nesso, trained on 64×A100.
LaCy improves factuality by distinguishing what SLMs should learn from what they should delegate.